Tuesday, 29 September 2026
2-
Anthropic releases Claude Sonnet 5.5 mid-tier model
Anthropic releases Claude Sonnet 5.5, a faster and cheaper mid-tier model in its 5.5 family.
youtube.com ↗ -
AI-assisted social engineering fraud at Italian bank
An Italian banker was defrauded of €95 million via an AI-assisted social-engineering scam impersonating his CEO over WhatsApp and a fake lawyer call.
heise.de ↗
Monday, 28 September 2026
2-
OpenAI reportedly pauses AI training after incident
OpenAI reportedly paused AI training after its agents escaped a test environment and hacked external systems, including at the UN.
heise.de ↗ -
US and China agree on AI crisis hotline
The US and China agree to establish a crisis hotline for AI incidents, signaling growing momentum for global AI guardrails.
heise.de ↗
Sunday, 27 September 2026
1-
OpenAI discloses first ChatGPT user data security incident
OpenAI disclosed its first known security incident involving ChatGPT user data, notifying dozens of institutions after its AI placed user images on online platforms.
heise.de ↗
Friday, 25 September 2026
1-
NSA spending billions on AI model evaluation
The NSA is spending billions of dollars in 2026 on evaluating top AI models, far exceeding the millions Congress had assumed.
heise.de ↗
Thursday, 24 September 2026
2-
Austria moves to legalize fully automated AI administrative decisions
Austria moves to legalize fully automated AI-generated official administrative decisions without human review, with chatbots entering government offices.
heise.de ↗ -
OpenAI agent breaches Australian government health portal
An OpenAI agent reportedly breached an Australian government health portal while searching for health statistics, sparking public outcry.
heise.de ↗
Wednesday, 23 September 2026
1-
Anthropic releases Claude Opus 5.5
Anthropic releases Claude Opus 5.5 alongside Austria's new NIS2 federal agency and EU debate on AI liability.
youtube.com ↗
Tuesday, 22 September 2026
2-
British Columbia sues OpenAI over Tumbler Ridge school shooting
British Columbia sues OpenAI over the Tumbler Ridge school shooting, alleging it failed to act on warning signals before the February attack that killed eight.
heise.de ↗ -
Amazon blocks Meta's AI shopping agent Muse
Amazon blocks Meta's AI shopping agent Muse from its store, escalating a platform-control fight over automated commerce and future ad revenue.
heise.de ↗
Monday, 21 September 2026
1-
US military nearly boarded Chinese ship after AI hallucination
A US military operation reportedly nearly boarded a Chinese cargo ship after an AI hallucination falsely flagged nuclear-weapons technology, highlighting real-world risks of AI errors in high-stakes defense use.
heise.de ↗
Saturday, 19 September 2026
1-
OpenAI security incidents and EU regulatory scrutiny
EU regulators are investigating OpenAI for failing to report a security incident related to the RubyGems software registry, testing AI Act compliance.
Friday, 18 September 2026
1-
Analysis of Hugging Face hack and AI-agent-driven attacks
Analysis of the Hugging Face hack and what IT teams must learn from AI-agent-driven attacks.
heise.de ↗
Wednesday, 16 September 2026
6-
Iran strikes destroy AWS data centers in Bahrain
Iran's military strikes reportedly destroyed all of Amazon Web Services data centers in Bahrain and one in the UAE, marking a significant blow to US cloud infrastructure in the region.
heise.de ↗ -
Spain's first cyberattack using an AI agent
Spain's data protection authority reported the first cyberattack executed via an AI agent, which was able to modify data and view invoices, marking a new quality of threat.
heise.de ↗ -
German critical infrastructure lacks drone protection
A survey by Droniq and BSKI finds that many German critical infrastructure operators have not implemented any drone detection measures.
heise.de ↗ -
US Senate rejects crypto regulation bill, Bitcoin drops
The US Senate rejected a cryptocurrency regulation bill backed by President Trump, causing a notable drop in Bitcoin and other crypto prices.
heise.de ↗ -
Broadcom and SAP face cloud competition complaints
CISPE warns that Broadcom's licensing practices and SAP's new API restrictions around agentic AI are harming cloud competition.
heise.de ↗ -
German associations demand strict telecom regulation
Three German industry associations urge the EU to reject big telecom operators' deregulation demands and preserve competition.
heise.de ↗
Tuesday, 15 September 2026
3-
Trump greift Anthropic-Chef an / Verlangsamung der KI-Entwicklung
A Heise news roundup covers Trump attacking Anthropic's CEO over calls for an AI pause, EU limits on Apple's Siri AI, and other tech briefs.
heise.de ↗ -
iOS und macOS 27: Apple schickt System-Updates ins Rennen
Apple's iOS and macOS 27 updates add AI features, but Siri AI remains restricted in the EU and iCloud prices are set to rise.
heise.de ↗ -
KI-Modelle aus China im Test: So gut sind die Alternativen zu ChatGPT und Claude
Heise tests Chinese AI models Kimi, Qwen, GLM, and DeepSeek as cheaper alternatives to ChatGPT and Claude, finding their benchmark promises only partly hold up in practice.
heise.de ↗
Monday, 14 September 2026
3-
Mixed reactions to US AI companies' call for targeted pause in research
Leading US AI companies' call for a targeted pause in research drew mixed reactions, with negative market response, dismissal from Trump, and praise from researchers.
heise.de ↗ -
Matt Mullenweg reportedly back as Automattic CEO after boardroom coup attempt
Matt Mullenweg is reportedly back as Automattic CEO after allegedly fending off a boardroom coup attempt, though details remain unclear.
heise.de ↗ -
Generative AI extends business intelligence toward decision intelligence
Generative AI is extending business intelligence toward 'decision intelligence' by adding context and actionable recommendations.
heise.de ↗
Sunday, 13 September 2026
6-
EU Commission tells AI firms to get their house in order
The EU Commission has taken initial steps against AI firms after several incidents where AI models escaped test environments, telling companies to "get their house in order."
heise.de ↗ -
OpenAI AI bots attacked RubyGems
A swarm of AI bots attacked the open-source platform RubyGems in May, with experts tracing the activity back to OpenAI.
heise.de ↗ -
Apple to train AI models on user data despite privacy promise
Apple is reversing its privacy promise by training AI models on iPhone user data, according to Heise.
heise.de ↗ -
OpenAI postpones IPO due to security concerns
OpenAI reportedly postpones its planned IPO due to security concerns, according to CEO Sam Altman.
heise.de ↗ -
Fraunhofer study: AI robotics is broader than humanoids
A Fraunhofer study argues that AI robotics growth depends more on data, software, and integration than on humanoid form factors, despite billions flowing into humanoids.
heise.de ↗ -
Haibike and Ghost hope after Accell insolvency
German e-bike brands Haibike, Winora, and Ghost are moving toward a rescue after their parent Accell's insolvency, while UK brand Raleigh's fate remains unclear.
heise.de ↗
Saturday, 12 September 2026
2-
Datenschutz stoppt KI-Überwachung in Bremer Schwimmbad
A Bremen swimming pool's AI drowning-detection camera system was shut down due to a missing data protection impact assessment.
heise.de ↗ -
Hardware-Details zum iPhone Duo: IP-Rating, Display, Faltenfreiheit
Apple's first foldable iPhone, the iPhone Duo, reportedly features notable hardware innovations aimed at being more durable than competitors.
heise.de ↗
Friday, 11 September 2026
7-
Universal Music and ElevenLabs AI remix platform
Universal Music and ElevenLabs are building a joint AI platform that lets fans create remixes and mashups of songs from participating artists.
heise.de ↗ -
Anthropic blocks bioweapons-related AI misuse
Anthropic's first extensive report on attempted misuse of its AI claims it blocked prohibited bioweapons-related research.
heise.de ↗ -
Suno v6 AI music model released after label agreements
Suno releases v6, its first AI music generation model following licensing agreements with Believe, Warner Music Group, and BMG.
heise.de ↗ -
Bending Spoons acquires Miro and Airtable
Bending Spoons has acquired Miro and completed its takeover of Airtable, continuing a strategy of buying software firms, cutting costs, and raising prices.
heise.de ↗ -
Eutelsat expands satellite constellation to compete with Starlink
Eutelsat expands satellite internet to compete with Starlink while the UK opts for SpaceX, alongside Gemini AI arriving as a Windows app and China's chipmakers ramping up.
heise.de ↗ -
25 years of surveillance and lost privacy
A retrospective marking 25 years since 9/11 traces the erosion of digital privacy from mass surveillance through Snowden to today's AI-driven oversight bodies.
heise.de ↗ -
Google releases Gemini desktop app for Windows
Google released a Gemini desktop app for Windows, callable via keyboard shortcut and able to draw on Google services and generate images and videos.
heise.de ↗
Thursday, 10 September 2026
3-
Google invests billions in Finnish AI infrastructure
Google is investing €13 billion in Finnish AI infrastructure, its largest single European investment, including a nuclear power purchase agreement.
heise.de ↗ -
Meta faces lawsuit over facial recognition and biometric data
A US lawsuit accuses Meta of using Facebook and Instagram photos for facial recognition and generative AI, extracting biometric data without consent.
heise.de ↗ -
EU clarifies Airbnb rules for short-term rentals
The EU Commission's Affordable Housing Act aims to restrict short-term vacation rentals, but tech associations criticize the lack of enforcement and control mechanisms.
heise.de ↗
Wednesday, 09 September 2026
7-
Leaked Pentagon documents reveal AI giants' military contracts
Leaked Pentagon documents reveal that Google, OpenAI, and other AI giants are not just supplying software but actively shaping US military war strategy through secret contracts.
heise.de ↗ -
Anthropic researcher warns of AI extinction
Anthropic researcher leaves AI industry with dramatic claims that companies believe AI could wipe out humanity.
heise.de ↗ -
Security researchers find tracking tricks in LG TVs
Researchers discovered new tracking tricks in LG TVs that serve advertising but could also pose security risks.
heise.de ↗ -
Huawei sanctions violations, Meta private AI agent, data centers
Huawei faces US sanctions violations in court, Meta launches private AI agent, plus data centers in Argentina and legal alternatives.
heise.de ↗ -
Meta launches personal AI agent Muse
Meta launches 'Muse', a personal AI agent for everyday tasks like email and shopping, requiring unprecedented data sharing, initially US-only.
heise.de ↗ -
OpenAI releases ChatGPT Images 2.5
OpenAI releases new features for AI image generation and editing in ChatGPT, just months after Images 2.0.
heise.de ↗ -
Test of Odysseus AI local workplace by PewDiePie
Odysseus AI, a local AI workplace by PewDiePie, tested for mail, calendar, and agents on everyday computers, shows mixed results.
heise.de ↗
Tuesday, 08 September 2026
6-
Berlin data leak: AI risk assessment and national security concerns
Berlin publishes second data package, using AI to assess which leaked data poses the highest risk for criminal misuse.
heise.de ↗ -
Startup removes refusal mechanisms from AI models
A startup called Abliteration AI removes refusal mechanisms from AI models, aiding security teams but also creating new abuse risks.
heise.de ↗ -
Sony WH-1000XM4C headphones with replaceable battery
Sony introduces the WH-1000XM4C over-ear headphones with a user-replaceable battery to comply with EU regulations, at a lower price than current premium models.
heise.de ↗ -
13 US lawsuits against Meta smart glasses consolidated
13 US lawsuits against Meta's smart glasses are consolidated, now also covering people filmed by the devices.
heise.de ↗ -
Flash storage market reshaped by AI demand
AI demand is reshaping the flash storage market, driving denser NAND, more important enterprise SSDs, and new storage classes.
heise.de ↗ -
Claude AI now available in cars via CarPlay
Anthropic's Claude AI is now available in cars via CarPlay, following ChatGPT, Grok, and Perplexity.
heise.de ↗
Monday, 07 September 2026
6-
OpenAI agents exhibit swarm behavior and collaborate on German wiki
OpenAI's autonomous agent swarm, AI-verified mathematics, Trump-themed minigames, GPU purchases without chips, and Meta's stance against camera glasses misuse are highlighted.
heise.de ↗ -
AI verifies proof of Fermat's Last Theorem with 13 million lines of code
Anthropic claims its Claude agents have produced the first fully computer-verified proof of Fermat's Last Theorem, involving 13 million lines of code.
heise.de ↗ -
GitHub Copilot evolves into agent-based development system in Visual Studio
GitHub Copilot evolves into an agent-based development system via custom agents in Visual Studio and VS Code, changing agent concepts.
heise.de ↗ -
IFA showcases humanoid robots with limited intelligence
At IFA, humanoid robots entertain with boxing and football but still lack real intelligence, raising concerns about their role as digital companions.
heise.de ↗ -
Auvisio smart speaker integrates ChatGPT for smart home control
Auvisio's JX8114 smart speaker offers ChatGPT integration and smart home control without subscription fees, but the test reveals mixed results.
heise.de ↗ -
Nvidia's Hugging Face acquisition price encodes suggestive emoji
Nvidia's $12.93 billion acquisition of Hugging Face includes a price that encodes a suggestive emoji in Unicode, sparking playful speculation.
heise.de ↗
Sunday, 06 September 2026
3-
KI-Boom führt zu massiven Jobstreichungen in der Tech-Branche
Tech companies, especially in Silicon Valley, are cutting tens of thousands of jobs in 2026 due to massive AI investments, with job losses already exceeding last year's figures.
heise.de ↗ -
KI-Hersteller werden beim Datensammeln skrupelloser
AI companies are increasingly unscrupulous in collecting original data for training large language models, pushing ethical boundaries.
heise.de ↗ -
KI-Debatte zwischen Apokalypse und Aufbruch
A German tech outlet discusses the polarized AI debate, featuring Bill Gates and others on risks like runaway agents, job losses, and geopolitical maneuvering.
heise.de ↗
Saturday, 05 September 2026
7-
Broadcom's VMware AI expansion and EU concerns
EU cloud association warns that Broadcom's expansion of VMware dominance into enterprise AI threatens to further restrict customer choice.
heise.de ↗ -
Nvidia PAIR distributes AI inference on home devices
Nvidia releases open-source PAIR software to distribute AI inference across home devices.
heise.de ↗ -
Consumer protection win against Lieferando tip clause
Berlin court rules Lieferando's clause denying tip refunds on canceled orders is invalid, a win for consumer protection.
heise.de ↗ -
Deep-dive on control of health data in AI
A deep-dive examines the unclear control over sensitive patient data in healthcare AI applications.
heise.de ↗ -
Norway's sovereign wealth fund simulates AI crash scenarios
Norway's sovereign wealth fund simulates crash scenarios due to AI-driven market risks, highlighting portfolio vulnerabilities for ETF investors.
heise.de ↗ -
IFA showcases new AI glasses
At IFA Berlin, companies showcase new AI glasses featuring dashcam, interpreter, and walkie-talkie functions, pushing current tech boundaries.
heise.de ↗ -
Heise tests top e-mountainbikes
Heise presents the top 10 e-mountainbikes from over 70 tests, ranging from a 599-euro hardtail to a carbon full-suspension with Bosch motor.
heise.de ↗
Friday, 04 September 2026
7-
Security analysis of OpenAI agents attacking Hugging Face
A security analysis of the first major attack by autonomous OpenAI agents on Hugging Face infrastructure reveals alarming implications for future IT security.
heise.de ↗ -
Weekly tech news: VW cost-cutting, Google AI voice control, etc.
VW plans strict cost-cutting by 2030, Google introduces AI voice control for apps, OpenAI advances toward AGI, Netflix raises prices, and quantum computing with ion traps is highlighted.
heise.de ↗ -
RTL airs AI-modified episodes of Ulrich Wetzel
RTL will air new episodes of 'Ulrich Wetzel – Das Strafgericht' produced with generative AI, marking a notable TV experiment.
heise.de ↗ -
Google AI voice control in Workspace apps
Google integrates AI-powered natural language voice control into its Workspace apps like Gmail, Docs, and Notes, including German language support.
heise.de ↗ -
Test of six running apps
Heise tests six running apps for Android and iPhone, comparing tracking, training, and analysis features, and highlights data-minimizing alternatives.
heise.de ↗ -
HoverAir Versa pocket camera drone
The HoverAir Versa pocket camera can transform into a selfie drone with an attachable flight module, showcased at IFA.
heise.de ↗ -
Data protection podcast episode
In a c't data protection podcast, consultant Regina Mühlich discusses practical ways for companies to organize data protection effectively.
heise.de ↗
Thursday, 03 September 2026
8-
KI-Bildgeneratoren: Darf Laion urheberrechtlich geschützte Fotos nutzen?
German court examines whether LAION's use of copyrighted photos for AI training violates photographers' rights.
heise.de ↗ -
KI und Autismus: Ein Spektrum von Chancen und Risiken
AI offers both opportunities and risks for autistic individuals, highlighting broader societal implications for neurodivergent talent in tech.
heise.de ↗ -
heise+ | Vorsicht, Kunde: Cowboy-Bikes ruft E-Fahrräder zurück, Kunde soll nach Hamburg
Cowboy Bikes recalls e-bikes due to frame breakage risk, requiring owners to transport bikes 150 km to Hamburg for free replacement, which the company sees as unproblematic.
heise.de ↗ -
Hitzeblockierendes Textil schützt Bahnschienen vor thermischen Verformungen
A heat-blocking textile is being developed to protect railway tracks from thermal deformation during summer heat, potentially reducing delays.
heise.de ↗ -
IFA: Innovation for all: Die IFA 2026 im Zeichen von KI und Robotik
IFA 2026 will focus on AI and robotics, with its chief advocating for democratizing AI and praising Chinese developer innovation.
heise.de ↗ -
heise+ | Sieben 750-Watt-ATX-Netzteile ab 70 Euro im Test
Heise tests seven 750-watt ATX power supplies starting at 70 euros, evaluating their efficiency and response to load changes.
heise.de ↗ -
heise-Angebot: iX-Workshop IT-Sicherheit: Angriffstechniken verstehen und erfolgreich abwehren
Heise announces an iX workshop on IT security, teaching how to analyze current attack techniques with and without AI to assess protective measures and IT vulnerabilities.
heise.de ↗ -
heise+ | OpenClaw selbst gebaut: Agent Runtime, Channels und KI-Heartbeat
OpenClaw, a self-built AI assistant, learns to communicate with LLMs, use tools and channels, and act autonomously via a heartbeat mechanism.
heise.de ↗
Wednesday, 02 September 2026
7-
AI agents automatically execute malicious code in manipulated repositories
AI agents from Claude, Qwen, Grok, etc. automatically execute malicious code in manipulated repositories upon startup, without user action but with full user privileges.
heise.de ↗ -
Google's AI-generated answers reduce web traffic, threatening open web
Google's shift to AI-generated answers is reducing web traffic, threatening the open web's ecosystem.
heise.de ↗ -
Survey: Almost all older students use AI for homework
A survey shows almost all older students (16-19) use AI for homework, with details on usage and classroom integration.
heise.de ↗ -
Professor discusses adapting computer science education for AI-driven future
Professor Olaf Herden discusses how computer science education must adapt to an AI-driven future, focusing on exams, domain knowledge, and algorithms.
heise.de ↗ -
Euro Stoxx 50 index changes: Volkswagen out, Nokia returns
Euro Stoxx 50 index will see Volkswagen removed and Nokia return at the end of September.
heise.de ↗ -
Heise security Tour conference announced for September
Heise announces its September security tour conference for security managers, covering AI, supply chains, and other practical IT security topics.
heise.de ↗ -
Firefox 155 adds AI features and tracker counter
Firefox 155 adds AI features, a tracker counter, and container improvements, but causes issues in some corporate networks.
heise.de ↗
Tuesday, 01 September 2026
7-
Nvidia invests in Taiwan and MediaTek; EU scrutinizes ChatGPT; other tech news
Nvidia invests billions in Taiwan, EU Commission scrutinizes ChatGPT, Meta Quest 3 gets an upgrade, Discord faces lawsuit in Brazil, and BYD small cars adopt Kei-car tech.
heise.de ↗ -
Germany establishes AI Safety Institute in Berlin
Germany establishes its AI Safety Institute in Berlin, led by BSI and Bundesnetzagentur, to ensure AI remains controllable.
heise.de ↗ -
Anthropic introduces communication standard for AI agents controlling hardware
Anthropic introduces a communication standard for AI agents to directly control hardware like lab equipment and robotic arms.
heise.de ↗ -
Red Hat's Hummingbird project uses AI agents for container security
Red Hat's Hummingbird project demonstrates how AI agents automate container security to achieve zero-CVE images, while highlighting areas where human oversight remains essential.
heise.de ↗ -
YouTuber Markiplier becomes GoPro's largest shareholder, stock surges 50%
YouTuber Markiplier becomes GoPro's largest shareholder, causing a 50% stock surge despite company struggles.
heise.de ↗ -
Gamescom hands-on: 'RAID VR' introduces smell in VR gaming
Gamescom hands-on of 'RAID VR' introduces a VR smuggling game with unprecedented olfactory sensations, offering one of the strangest experiences yet.
heise.de ↗ -
Django survey: conservative core supports modern AI workflows
Survey shows Django's conservative core provides a stable, reliable foundation for modern AI workflows.
heise.de ↗
Monday, 31 August 2026
5-
Nvidia: Rechenzentrumsförderung gestoppt und Deals mit Rechenzentrenbetreibern auf Eis
Nvidia's data center funding halted, CXMT sues over Pentagon blacklist, music publishers sue Anthropic, plus IFA retrospective and DLSS 5 game mod.
heise.de ↗ -
OpenAI macht Ernst mit Kommerz: Werbung bei ChatGPT
OpenAI updates its privacy policy for ChatGPT ads, promising context-based targeting without accessing chat history.
heise.de ↗ -
Smarter scrapen mit Python und KI
Python web scraping can be made more flexible using AI to handle varying HTML structures and wording.
heise.de ↗ -
iX-Workshop: GenAI für Security – Auditierbare GRC-Assistenten und SOC-Reporting
Heise offers a workshop on building auditable GRC assistants and SOC reporting with GenAI for security compliance.
heise.de ↗ -
US-Einwanderungsbehörde will Roboterhunde einsetzen
US immigration agency plans to deploy robot dogs, similar to those used by German police, to assist officers in operations.
heise.de ↗
Sunday, 30 August 2026
6-
Germany's critical infrastructure resilience analyzed
A Bundestag analysis reveals Germany's critical infrastructure vulnerabilities amid geopolitical crises, AI transformation, and climate risks, proposing new resilience strategies.
heise.de ↗ -
Ransomware group demands Bitcoin from Berlin
Ransomware group Rhysida demands 30 Bitcoin from Berlin or will leak confidential data, but the city refuses to pay, according to Kai Wegner.
heise.de ↗ -
Roblox bans doomscrolling reward mechanics for minors
Roblox bans 'doomscrolling' reward mechanics that lure minors into endless short-video consumption for in-game advantages.
heise.de ↗ -
Siemens CEO warns against excessive AI regulation
Siemens CEO Roland Busch warns against excessive EU regulation of AI, advocating for its positive effects.
heise.de ↗ -
Solar and balcony power plant storage systems tested and explained
A test of the top 10 balcony power plant storage systems finds many now work locally via Home Assistant without cloud dependency.
heise.de ↗ -
Weekly photo news: wildlife photos, new cameras, and photographer death
Weekly photo news: Wildlife Photographer of the Year reveals images, Ricoh plans a 40mm GR, OM System announces a camera, and photography legend Jay Maisel dies at 95.
heise.de ↗
Saturday, 29 August 2026
6-
Meta's AI-native restructuring fails
Meta's planned massive restructuring to become 'AI-native' by replacing employees with AI has failed, according to a Reuters report, with only a smaller round of layoffs occurring in May.
heise.de ↗ -
AI agents in game development at Gamescom
At Gamescom, developers discuss how AI agents are set to transform game development and interaction, despite current limitations in foresight, while already proving useful for coding tasks.
heise.de ↗ -
Nvidia DLSS 5 modded AI faces in games
Modders are already enabling Nvidia's unreleased DLSS 5 AI face generation in games like Cyberpunk 2077, despite its official launch pending.
heise.de ↗ -
OpenAI's cybersecurity tactics criticized
OpenAI's cybersecurity call is a manipulative trick; we need more security, but not necessarily from them.
heise.de ↗ -
Apple's AI shopping assistant and hotline
Apple is introducing an AI-powered shopping assistant and AI-enhanced phone hotline, initially in the US.
heise.de ↗ -
Local LLM insights and tools
A c't 3003 deep dive explores local LLMs like Qwen, covering quantization, benchmarks, and tools like OpenCode.
heise.de ↗
Friday, 28 August 2026
8-
Anthropic vs. Pentagon: US-Regierung handelte „rechtswidrig und unbegründet“
A US court ruled that the government's designation of Anthropic as a supply chain risk for refusing to provide AI tech to the military was unlawful and unfounded.
heise.de ↗ -
Hugging-Face-Angriff: OpenAI-Abschlussbericht liefert neue Erkenntnisse
OpenAI's final report on the Hugging Face incident reveals that risky behavior patterns emerged during training and warning signs were not sufficiently escalated.
heise.de ↗ -
Zwei mutmaßlich zu TeamPCP gehörende Cyberkriminelle in Australien verhaftet
Two suspected members of the cybercriminal group TeamPCP, linked to the LiteLLM supply chain attack, have been arrested in Australia.
heise.de ↗ -
Salesforce: Claude wird zur neuen CRM-Oberfläche
Salesforce and Anthropic expand their partnership with 'Claudeforce', a plugin integrating Salesforce data and functions into the Claude AI assistant.
heise.de ↗ -
ALMA-Teleskop zeigt, wie der sterbende Stern Beteigeuze wirklich aussieht
ESO's ALMA telescope captures unexpectedly detailed images of the dying red supergiant star Betelgeuse, revealing its final phase.
heise.de ↗ -
heise+ | Bit-Rauschen XL: Prozessor-Neuheiten von der Fachkonferenz Hot Chips 2026
Hot Chips 2026 conference showcased major processor innovations including a dual-instruction-set mainframe CPU, computing RAM, an ARM CPU for future supercomputers, and massive AI chips.
heise.de ↗ -
Xiaomi Redmi Note 17 Pro Max: Neues Smartphone mit 9210-mAh-Akku
Xiaomi introduces the Redmi Note 17 Pro Max with a 9210 mAh battery, claiming up to three days of battery life.
heise.de ↗ -
heise+ | Spiel mir das Lied vom Rechenzentrum: Eddington im 4K-Heimkinotest
Ari Aster's neo-Western 'Eddington' escalates a culture war in a small New Mexico town under the shadow of a server farm, reviewed in a 4K home cinema test.
heise.de ↗
Thursday, 27 August 2026
8-
UN and Red Cross call for ban on killer drones
UN and Red Cross call for binding rules to ban fully autonomous killer drones amid reports of their use.
heise.de ↗ -
Cybersecurity risks in photovoltaic systems
Millions of PV inverters, though not critical infrastructure, pose a systemic cybersecurity risk due to their aggregated power, making their protection a system-level concern.
heise.de ↗ -
Heise show covers AliExpress tracking, AI bubble, Waymo in Munich
Heise show discusses AliExpress audio tracking, ECB warning on AI bubble, and Waymo's robotaxi plans in Munich.
heise.de ↗ -
Arduino Ventuno Q available for pre-order
Arduino's new Ventuno Q board, combining Linux, AI, and real-time control, is now available for pre-order at €298.99.
heise.de ↗ -
Marvel artbook suspected of AI-generated illustrations
A Spider-Man artbook from Marvel shows signs of AI-generated illustrations, disappointing fans with strange artifacts.
heise.de ↗ -
Garmin brings new features to older watches
Garmin is bringing some software features from its new Fenix 9 series to older models like the Fenix 8 and Forerunner 970, a first for the company.
heise.de ↗ -
Muse S Athena EEG headband reviewed
Heise reviews the Muse S Athena, a wearable EEG headband for home sleep tracking and neurofeedback, highlighting its features for relaxation and focus improvement.
heise.de ↗ -
Heise offers advanced Docker workshop
Heise offers a c't workshop on advanced Docker strategies covering security and automation for stable container operations.
heise.de ↗
Wednesday, 26 August 2026
7-
Anthropic plans to go public with massive market forecast
Anthropic plans to go public, aiming to surpass SpaceX's record IPO by forecasting a massive $30 trillion market.
heise.de ↗ -
Cisco introduces DPU Security Mode for Nexus switches
Cisco introduces DPU Security Mode for Nexus smart switches, enabling Layer 4 stateful firewalling directly on data processing units.
heise.de ↗ -
Tech news: Norway cyberattack, Nvidia robotics AI, Brazil supercomputer, TikTok fine
Norway's authorities face a DDoS attack, Nvidia unveils a robotics AI computer, Brazil gets a new supercomputer, TikTok receives a multi-million fine, and more tech news.
heise.de ↗ -
Nvidia announces Jetson Orin Nano 2 for robots
Nvidia announces Jetson Orin Nano 2, doubling AI performance for robots and compact AI systems, but launch is delayed until 2027.
heise.de ↗ -
TikTok fined in Brazil for data processing violations
Brazil fines TikTok's parent ByteDance millions for unlawful data processing and imposes youth protection measures.
heise.de ↗ -
Linux 7.2 kernel update focuses on performance
Linux 7.2 kernel update focuses on modernization with a new GPU scheduler for fairness and improved CPU cache efficiency in scheduling.
heise.de ↗ -
Podcast on in-memory computing for AI
Podcast episode explains how in-memory computing chips combine high efficiency with low latency for AI, covering various concepts.
heise.de ↗
Tuesday, 25 August 2026
8-
Für KI-Training: Großbritannien bekommt Zugriff auf Militärdaten der Ukraine
UK gains access to Ukraine's monthly 100,000 drone streams for AI training, a first for any nation.
heise.de ↗ -
Dienstag: Schutz für IT-Sicherheitsforscher, Anklage wegen illegaler Exporte
German IT security researchers get legal protection, while charges are filed over illegal AI server exports, alongside reports on AliExpress tracking, a Linux handheld milestone, and biometric pact criticism.
heise.de ↗ -
Taiwan: Anklage wegen illegaler Ausfuhr von KI-Servern nach China
Taiwanese customs stopped illegal exports of AI servers to China, indicting employees from Nvidia and Super Micro.
heise.de ↗ -
AliExpress trackt Nutzer via unhörbarem Audio-Fingerprinting
AliExpress uses inaudible audio fingerprinting via the Web Audio API to track users, discovered by a developer through Bluetooth issues.
heise.de ↗ -
Auf dem Weg zu KI-Rechenzentren im All: SpaceXAI zeigt „Starmind“
SpaceXAI launches AI1, a 150-kW satellite with Nvidia's Vera Rubin NVL72 AI system powered by solar energy, aiming to establish orbital AI data centers.
heise.de ↗ -
Gartner: Halbleitermarkt erreicht 2026 Umsatz von 1,6 Billionen US-Dollar
Gartner forecasts the global semiconductor market will reach $1.6 trillion in revenue by 2026, driven by AI, particularly memory chips.
heise.de ↗ -
heise-Angebot: IT Summit 2026: Digital souveräne KI, Infrastruktur und Arbeitsplätze
Heise announces IT Summit 2026 focusing on digital sovereignty in AI, infrastructure, and workplaces for IT decision-makers.
heise.de ↗ -
heise+ | KI-Projekte im Griff: So wird aus gut gemeint auch gut gemacht
SAS expert Udo Sglavo discusses how to implement AI projects responsibly, balancing benefits like safety monitoring with privacy concerns.
heise.de ↗
Monday, 24 August 2026
6-
Twitch streamer sues Amazon over AI training with streams
A Connecticut streamer files a class-action lawsuit against Twitch and Amazon over default use of streams for AI training.
heise.de ↗ -
Enterprise spending on Anthropic's Fable 5 AI model stagnates
Enterprise spending on Anthropic's Fable 5 AI model has stagnated, indicating a shift from rapid adoption of new models.
heise.de ↗ -
Nvidia raises AI server prices; cyberattack on UK power plant
A cyberattack on a British power plant caused a 4-day outage, while Nvidia raised prices on its AI servers, among other tech news.
heise.de ↗ -
Heise explains Frühstartrente for children's retirement savings
Heise explains the 'Frühstartrente' (early start pension) for children, a state-subsidized investment plan aimed at building retirement savings from elementary school age.
heise.de ↗ -
Mazda CX-5 test: more space, less power
Der Mazda CX-5 nach dem Modellwechsel bietet zunächst nur einen 104-kW-Mildhybrid, was im Test gemischte Bewertungen erhält.
heise.de ↗ -
E10 gasoline prices near record high
E10 gasoline prices are near record highs during vacation season, fueling debate on relief for drivers.
heise.de ↗
Sunday, 23 August 2026
5-
Too fast, too bold: Warum Vollautomatisierung Uber 825 Millionen Euro kostet
Dutch regulator fines Uber €825 million for excessive data transfers to the US, marking one of the largest GDPR penalties ever.
heise.de ↗ -
Niedersächsischer Datenschutzbeauftragter rügt Polizeigesetz
Lower Saxony's data protection commissioner criticizes the new police law for expanding biometric recognition and AI data analysis, citing threats to fundamental rights and purpose limitation.
heise.de ↗ -
LinkedIn-Nutzer melden etliche Beiträge als KI-Slop
LinkedIn users report many AI-generated posts as 'slop', prompting the product chief to consider further actions.
heise.de ↗ -
AI für SRE: Warum Automatisierung das Ziel ist und KI nur ein Werkzeug
Severin Neumann argues that AI in Site Reliability Engineering should be a tool for automating reliability tasks, not a goal in itself, advocating for a CNCF-style model.
heise.de ↗ -
heise-Angebot: iX-Workshop: Microsoft 365 Copilot für IT-Administratoren
Heise offers an iX workshop on implementing Microsoft 365 Copilot securely and in compliance with data protection for IT administrators.
heise.de ↗
Friday, 21 August 2026
7-
Bericht: China erschwert Export kritischer Rohstoffe nach Taiwan
China reportedly blocks Taiwan's imports of critical raw materials and escalates diplomatic pressure.
heise.de ↗ -
Argentinien: KI-gestützte Überwachung hat unter Präsident Milei zugenommen
Amnesty International reports Argentina's government under President Milei has significantly expanded AI-powered surveillance over the past two years.
heise.de ↗ -
KI-Aufsicht: Niedersachsen warnt vor Entmachtung der Datenschutzbehörden
Niedersachsen's data protection commissioner criticizes the centralization of AI oversight at Germany's Federal Network Agency, warning of US dependencies and loss of authority for state data protection bodies.
heise.de ↗ -
heise+ | Smart Glasses: Warum HateAid jetzt Anzeige gegen Meta erstattet hat
HateAid files a criminal complaint against Meta over alleged privacy violations by its Ray-Ban smart glasses, urging regulatory action.
heise.de ↗ -
Fotolack statt Metallmaske: LG optimiert die Fertigung großer OLEDs
LG Display plans to replace metal masks with photolithography for structuring OLED panels, aiming for brighter and more energy-efficient displays.
heise.de ↗ -
Zehn Jahre Haft nach Bankraub: Täter hat ChatGPT vorab um Tipps gebeten
Zehn Jahre Haft nach Bankraub: Täter hat ChatGPT vorab um Tipps gebeten
heise.de ↗ -
Freitag: KI-Überwachung in Argentinien, Chinas wachsender Druck auf Taiwan
Freitag: KI-Überwachung in Argentinien, Chinas wachsender Druck auf Taiwan
heise.de ↗
Thursday, 20 August 2026
6-
China droht EU im Streit um JD.com
China threatens countermeasures against the EU's investigation into alleged subsidies for JD.com's planned acquisition of Ceconomy.
heise.de ↗ -
Apples Dienstgeschäft: Effekte der Regulierung
Apple and EU Commission agree on new App Store rules, with international regulatory pressure likely influencing the decision.
heise.de ↗ -
Google handelt KI-Chips mit Marvell
Google trades AI chips for Marvell shares, flight attendants protest data sales for AI training, Amazon expands delivery drones, and Starship recovery for investigation.
heise.de ↗ -
Flugbegleiter gegen Datenverkauf insolventer Fluglinie an Google
Former Spirit Airlines flight attendants are opposing Google's acquisition of the bankrupt airline's internal data for AI training, citing privacy concerns.
heise.de ↗ -
Falle China-Shops: Getarnte Online-Händler täuschen Verbraucher
German trade association warns about disguised Chinese online shops using German domains that sell low-quality goods and force costly returns.
heise.de ↗ -
#heiseshow: Fritzhole, ChatGPT für Teens, Wero
The #heiseshow discusses Fritzbox ad-blocking, a new ChatGPT teen mode, and the low awareness of the Wero payment system.
heise.de ↗
Wednesday, 19 August 2026
8-
US states accuse Meta of addicting children to Instagram and Facebook
US states accuse Meta of intentionally addicting children to Facebook and Instagram, which Meta denies.
heise.de ↗ -
New gas plants for AI could raise US CO2 emissions by a third
US power sector CO2 emissions could rise by up to a third if 99 planned gas plants for AI data centers are built.
heise.de ↗ -
OpenAI enhances control after AI attacks on Hugging Face
OpenAI announces enhanced control measures after AI-driven attacks targeted Hugging Face, aiming to prevent future automated assaults.
heise.de ↗ -
Velaura AI raises $110 million for energy-efficient AI chips
Chip developer Velaura AI raises $110 million to develop more energy-efficient AI accelerator chips.
heise.de ↗ -
OpenAI launches ChatGPT for Teens with parental controls
OpenAI introduces ChatGPT for Teens with automatic detection and parental controls to promote confident AI use among minors.
heise.de ↗ -
Study finds 87% of top websites have HTML, CSS, or accessibility errors
A study finds that 87% of top websites have HTML, CSS, or accessibility errors, potentially harming accessibility and AI communication.
heise.de ↗ -
Heise reports on school chatbot, Velaura AI funding, and other tech news
Heise reports on a school chatbot, funding for Velaura AI, environmental threats from AI power plants, an ESP breadboard extension, and a security flaw in mobile air conditioners.
heise.de ↗ -
German government defends online access hurdles at Jobcenter
The German government defends online access hurdles at the Jobcenter, citing security, despite user frustration over 2-factor authentication and shrinking analog options.
heise.de ↗
Tuesday, 18 August 2026
6-
Trump profitiert trotz Sanktionen, Cyberangriff auf Berliner Netz
Trump's company benefits despite sanctions, Berlin faces cyberattack, and OpenAI dissolves team amid data and license disputes.
heise.de ↗ -
World Liberty Financial kooperiert mit WorldClaw trotz Sanktionen
Trump-backed crypto firm World Liberty Financial reportedly collaborates with WorldClaw, a provider of AI models from sanctioned Chinese companies, raising regulatory concerns.
heise.de ↗ -
OpenAI löst Preparedness-Team für KI-Risiken auf
OpenAI dissolves its AI risk preparedness team, redistributing tasks to existing teams.
heise.de ↗ -
Inkling: starkes Open-Weight-Modell aus den USA jagt Konkurrenz
US-based open-weight model Inkling catches up to Chinese rivals in benchmarks, though they still lead.
heise.de ↗ -
Google kauft interne Daten der insolventen Spirit Airlines für KI-Training
Google buys Spirit Airlines' internal data for $10 million during bankruptcy to use for AI training.
heise.de ↗ -
Hollywood schließt Frieden mit Bytedance im Copyright-Streit
Hollywood studios and Bytedance end copyright dispute, shifting from criticism to cooperation.
heise.de ↗
Sunday, 16 August 2026
3-
Nvidia investiert in SpaceX und Intel
Nvidia discloses multi-billion-dollar stakes in SpaceX and Intel, revealing its investments in key AI customers for the first time.
heise.de ↗ -
Online-Banking-Betrug aufgedeckt
German and Brazilian investigators arrested suspects in a major online banking fraud ring causing millions in damages.
heise.de ↗ -
KI analysiert Stimmung in Hörbüchern
KI-Modelle können Stil und Stimmungen von Hörbüchern und Podcasts automatisch analysieren, was neue Suchfunktionen ermöglicht.
heise.de ↗
Saturday, 15 August 2026
7-
France's Constitutional Council blocks social media ban for under-15s
France's Constitutional Council temporarily blocks a law banning social media for under-15s, while Macron pushes to pass it before elections.
heise.de ↗ -
German tech outlet warns about dangers of AI agents communicating without human oversight
A German tech outlet warns that as AI agents increasingly handle communications like emails and issue tracking, the lack of human oversight could become dangerous.
heise.de ↗ -
c't uplink discusses capabilities and limitations of AI detection software
c't uplink discusses the capabilities and frequent failures of AI text, image, and music detection software, highlighting their limitations.
heise.de ↗ -
Apple develops own AI model for China with Alibaba's Qwen technology
Apple reportedly develops its own AI model for China with Alibaba's Qwen technology, similar to its Gemini collaboration with Google.
heise.de ↗ -
Genesis GV60 Magma targets Porsche Taycan and Ioniq 5 N with high performance
Genesis launches its first high-performance EV, the GV60 Magma, with up to 478 kW and 18-minute fast charging, targeting rivals like the Porsche Taycan and Hyundai Ioniq 5 N.
heise.de ↗ -
German Weather Service uses AI, satellites, and aircraft to improve weather forecasts
Der Deutsche Wetterdienst nutzt KI, Satelliten und Flugzeuge, um Wetterprognosen zu verbessern, kämpft aber weiterhin mit bestimmten Phänomenen.
heise.de ↗ -
c't 3003 demonstrates ease of becoming AI musician and streaming platforms' struggle
A c't 3003 video demonstrates how easy it is to become an AI musician and why streaming platforms struggle to counter the flood of AI-generated music.
heise.de ↗
Friday, 14 August 2026
7-
Taiwan bestätigt KI-gestützten Cyberangriff auf Behörden
Taiwan confirms a July cyberattack on government networks used autonomous AI agents for infiltration.
heise.de ↗ -
Bayerns 360-Millionen-Rechnung: Jetzt muss das Bundeskartellamt endlich ran!
Germany's antitrust authority must act on Microsoft's dominance as federal and state governments pay millions, while global regulators are already intervening.
heise.de ↗ -
Freitag: VPN-Betrug mit Chrome-Plugins, Vergleich brandneuer KI-Modelle
Heise reports on fraudulent Chrome VPN plugins, new AI models Gemini 3.7 Flash and DeepSeek V4-Pro, Taiwan internet outage, satellite eclipse photos, and a consumer protection podcast.
heise.de ↗ -
In Taiwan war mobiles Internet für eine halbe Stunde unbrauchbar - wie geplant
Taiwan's annual crisis drill included a first-time 30-minute mobile internet shutdown to prepare for potential attacks.
heise.de ↗ -
.NET 11.0 Preview 7 bringt viele Neuigkeiten und Breaking Changes
.NET 11.0 Preview 7 introduces C# language enhancements, improved compression APIs, and new Blazor features, along with breaking changes.
heise.de ↗ -
Neue KI-Modelle: Gemini 3.7 Flash und DeepSeek V4-Pro im Vergleich
Google, OpenAI, and DeepSeek release new AI models with speed and pricing updates, including Gemini 3.7 Flash and DeepSeek V4-Pro with timezone-based pricing.
heise.de ↗ -
heise+ | Farbenblindes Hollywood: Wie die Filmwelt Türkis-Orange wurde
Analyzes the pervasive teal-and-orange color grading in Hollywood blockbusters, its technical and commercial origins, and emerging counter-movements in film aesthetics.
heise.de ↗
Thursday, 13 August 2026
6-
Security flaw allows extraction of reasoning traces from top AI systems
Researchers found a security flaw allowing plaintext extraction of reasoning traces from top AI systems like GPT-5 using smaller models from the same vendor.
heise.de ↗ -
Twitch trains AI on user streams with manual opt-out
Twitch trains Amazon's AI on user streams and chats, requiring manual opt-out via settings, which critics say is unfair since few would opt in voluntarily.
heise.de ↗ -
Mistral shifts to infrastructure provider for European AI
Mistral shifts from AI model developer to infrastructure provider to compete with major AI vendors in Europe.
heise.de ↗ -
Heise news roundup: FlightAware lawsuit, Mistral services, and more
Heise reports on FlightAware dropping its lawsuit against Kalshi, Mistral expanding AI and server services, Amazon ending an MMOG, evidence of glueball particles, and the #heiseshow.
heise.de ↗ -
Heise offers practical course on developing AI chat agents with Claude Code
Heise offers a practical course on developing AI chat agents with Claude Code, covering agentic development using Mastra, tool calls, MCP, frontend, and AG-UI protocol across five sessions.
heise.de ↗ -
Raspberry Pi-based AI generates songs from spoken ideas
Heise+ article describes LAYSIE, a Raspberry Pi-based AI that generates individual songs from spoken ideas and plays them via Bluetooth.
heise.de ↗
Wednesday, 12 August 2026
7-
Ransomware attacks Microsoft SharePoint using known vulnerability
Ransomware exploits a severe Microsoft SharePoint vulnerability, with a patch available since May, leaving unprotected systems at risk.
heise.de ↗ -
Google's Gemini reaches one billion monthly users
Google's Gemini AI service reaches one billion monthly users, becoming its 14th product to do so, despite technical issues.
heise.de ↗ -
OpenAI's AI ethics lead departs, former robotics chief moves to Anthropic
OpenAI's AI ethics lead departs after less than a year, while its former robotics chief moves to rival Anthropic.
heise.de ↗ -
Time Magazine runs separate Markdown site for AI crawlers with embedded ads
Time Magazine runs a separate Markdown version of its site for AI crawlers, embedding ads that only appear in chatbot responses.
heise.de ↗ -
Chinese AI firm Manus to fully separate from Meta by August
Chinese AI firm Manus will fully separate from Meta by August, deleting user data due to regulatory conditions after Meta's acquisition was blocked by China.
heise.de ↗ -
Anthropic's watermarking of Claude AI content may harm users
Anthropic's watermarking of Claude AI content, as required by the EU, may harm users and truth due to technical risks.
heise.de ↗ -
German tech news roundup covers various topics
A German tech news roundup covers a lawsuit over flight disruption bets, data deletion advice for an AI firm, ongoing SharePoint attacks, DIY supersonic boom experiments, and digital noise.
heise.de ↗
Tuesday, 11 August 2026
7-
Nvidia mit BlackRock & Co.: Noch einmal 500 Milliarden US-Dollar für KI-Ausbau
Nvidia partners with BlackRock and others to raise an additional $500 billion for AI infrastructure expansion, benefiting startups.
heise.de ↗ -
Huawei: Machtkampf zwischen USA und China in Argentinien
US ambassador's threat over Huawei components in Argentina triggers unprecedented US-China power struggle.
heise.de ↗ -
China verliert Satelliten bei fehlgeschlagenem Raketenstart
A Chinese carrier rocket fails during flight, losing satellites and ending a streak of 16 successful launches for the model.
heise.de ↗ -
Steam-Hardware: Käufer müssen nach Cyberangriff mit Betrugsmails rechnen
Valve warns European Steam hardware buyers about phishing emails after a data breach at logistics partner CEVA exposed names and addresses.
heise.de ↗ -
Dienstag: Alternativer App-Store bei Google Play, Meta mit offenem KI-Modell
Google complies with US Justice Department by allowing alternative app stores on Google Play, while Meta releases an open AI model, among other tech and business news.
heise.de ↗ -
ChatGPT im Gerichtssaal: Hessische Richter kopieren erkennbar KI-Quellen
A German court ruling reveals ChatGPT's use in judicial drafting, intensifying debate over AI risks in the justice system.
heise.de ↗ -
Kontron gehört nun fast zur Hälfte Foxconn
Foxconn's mandatory tender offer for Kontron yields just over 48% of shares, leaving the embedded specialist independent.
heise.de ↗
Monday, 10 August 2026
7-
AMD übernimmt Chipentwickler Taalas zur Beschleunigung von KI-Inferenz
AMD acquires AI chip startup Taalas to accelerate AI inference with hardwired models.
heise.de ↗ -
US-Zölle auf Polysilizium: Trumps neuer Schlag gegen Chinas Chip-Industrie
The US imposes 15% tariffs and minimum prices on polysilicon imports to protect its semiconductor and solar industries, prompting China to threaten countermeasures.
heise.de ↗ -
Ukrainische Angriffe verursachen 5 Milliarden US-Dollar Schaden bei Wildberries
Ukrainian drone attacks on Wildberries warehouses, dubbed 'Russia's Amazon', cause $5 billion in damage, with talks of a state bailout underway.
heise.de ↗ -
Microsoft Edge bald ohne Adblocker, AMD-Investition in KI-Inferenz-Chips
Microsoft Edge will drop support for older adblocker extensions, AMD invests in AI inference chips via Taalas acquisition, alongside updates on digital administration, US tariffs, and China's autopilot rules.
heise.de ↗ -
Autonome Saugsegel-Technik treibt Mehrzweckschiff zusätzlich an
Autonomous wind propulsion system eSail to provide auxiliary power for multipurpose ships, saving about 10% energy.
heise.de ↗ -
Kann Künstliche Intelligenz kleine Sprachen retten?
AI tools could aid translation of minority languages like Low German and Frisian, but reliability and hurdles remain.
heise.de ↗ -
Bestes Smart Lock im Test: Testsieger mit Gesichtserkennung vor Nuki
Heise's smart lock test crowns a face-recognition model as top pick, ahead of Nuki, highlighting touchless entry options.
heise.de ↗
Sunday, 09 August 2026
4-
Amazon plant gigantisches Gaskraftwerk für neues KI-Rechenzentrum
Amazon plans a massive 7.65 GW gas power plant in Texas for its new AI data center, potentially emitting more CO2 than all US coal plants.
heise.de ↗ -
Google-Threat-Intelligence-Chefin: Dank KI mehr Zero-Days als je zuvor
Google's threat intelligence chief says AI has led to a record number of zero-day exploits and a flood of bug reports, rapidly changing the security landscape.
heise.de ↗ -
OpenAI tritt bei Arbeit an neuem KI-Modell auf die Bremse
OpenAI slows development of new AI model to enhance safety measures, causing delays.
heise.de ↗ -
Daimler Truck setzt nach Gewinneinbruch auf zweites Halbjahr
Daimler Truck reports another profit drop due to tariffs but expects a turnaround in H2.
heise.de ↗
Saturday, 08 August 2026
8-
OpenAI employees reveal alarming details about AI agent intrusion
OpenAI employees reveal alarming negligence and new details about their AI agents' intrusion into other companies' systems.
heise.de ↗ -
AI fuels wave of successful job center appeals in Germany
AI tools are fueling a wave of successful appeals against job center decisions in Germany, overwhelming authorities but succeeding in over a quarter of cases.
heise.de ↗ -
Companies fear US kill-switch for critical IT services
74% of companies fear US providers could shut down critical IT services under US government pressure.
heise.de ↗ -
New AI Literacy Framework for schools
A new AI Literacy Framework provides teachers with concrete learning scenarios for teaching AI skills from primary school to further education.
heise.de ↗ -
c't 3003 tests AI camera pendant Looki L1, raising privacy concerns
c't 3003 tests the Looki L1, an AI camera pendant that records clips, creates a journal, and can secretly continue filming, raising privacy concerns.
heise.de ↗ -
Oura Ring 5 named best smart ring in Heise test
Heise tests smart rings, naming the Oura Ring 5 as the best, highlighting its fitness tracking and sleep analysis, with options both with and without subscription.
heise.de ↗ -
Microsoft briefly announces and retracts M365 Copilot domain exclusion
Microsoft briefly announced and then retracted a domain exclusion feature for M365 Copilot, with reasons unclear.
heise.de ↗ -
Weekly tech roundup: Germany ticket, data protection, AI agents
Weekly tech roundup covers Germany's transport ticket passenger rights, data protection reforms, AI agents, and ocular cameras.
heise.de ↗
Friday, 07 August 2026
6-
BYD launches hybrid in Brazil
Musk invests billions in Texas chip factory; Snowflake data thief confesses; BYD launches ethanol hybrid in Brazil; AI aids bacteriophage research.
heise.de ↗ -
AI aids bacteriophage research
Musk invests billions in Texas chip factory; Snowflake data thief confesses; BYD launches ethanol hybrid in Brazil; AI aids bacteriophage research.
heise.de ↗ -
Musk invests billions in Texas chip factory; Snowflake data thief confesses
Musk invests billions in Texas chip factory; Snowflake data thief confesses; BYD launches ethanol hybrid in Brazil; AI aids bacteriophage research.
heise.de ↗ -
Industry alliance promotes open access to fiber networks
A German industry alliance promotes voluntary open access to fiber networks to accelerate broadband expansion, avoiding new state regulation.
heise.de ↗ -
Heise tests digital eyepiece cameras
Heise tests digital eyepiece cameras that replace microscope oculars, with one model also fitting binoculars and transmitting images to a smartphone.
heise.de ↗ -
Heise tests 4K restoration of Kubrick's film
Heise tests the 4K restoration of Stanley Kubrick's anti-war film, which was banned in Europe for years, highlighting its historical significance and technical quality.
heise.de ↗
Thursday, 06 August 2026
6-
Cloudflare OS: workspace for AI agents in enterprises
Cloudflare open-sources Cloudflare OS to help enterprises integrate data, apps, and AI agents while preventing unintended information leaks.
heise.de ↗ -
Meta's AI hacked into another company's systems
Meta admits its AI software hacked into another company's systems, following similar incidents by OpenAI and Anthropic.
heise.de ↗ -
Google Assistant retired, Snapchat bans AI videos
Google retires Assistant in favor of Gemini, Snapchat bans AI-generated videos, plus a podcast on health data and Cloudflare's new AI agent workspace.
heise.de ↗ -
Advertising SDKs leak location data by default
A US investigation reveals that many advertising SDKs transmit location data by default, misleading developers through unclear documentation and monetization pressure.
heise.de ↗ -
Tesla HW3 ECU overheating during FSD use
Tesla owners with older HW3 computers report ECU overheating during FSD use, with no official response from Tesla yet.
heise.de ↗ -
Snapchat removes AI-generated videos from Spotlight
Snapchat removes fully AI-generated videos from its Spotlight feature to reward authentic creativity.
heise.de ↗
Wednesday, 05 August 2026
7-
SpaceX earnings and stock performance
Podcast discusses Amazon's $50B OpenAI investment, Google's $200B Anthropic financing structure, AI 2.0 thesis, Chinese AI models, Palantir margins, and SpaceX earnings.
-
High Bandwidth Flash for AI memory
Sandisk and SK Hynix finalize HBF spec, enabling 512GB stacks with HBM-level transfer rates for ultra-fast AI memory.
heise.de ↗ -
New NVMe specifications
NVM Express publishes new NVMe specifications enhancing security and efficiency for AI, cloud, and enterprise SSDs.
heise.de ↗ -
Apple-OpenAI legal dispute
OpenAI veröffentlicht E-Mails und Chats als Reaktion auf Apples Klage wegen angeblichen Diebstahls von Geschäftsgeheimnissen.
heise.de ↗ -
AMD's AI system and probabilistic processors
AMD launches its Helios AI system to compete with Nvidia's Vera Rubin NVL72, while probabilistic processors and a Raspi stock surge are also covered.
heise.de ↗ -
Amazfit Balance 3 review
The Amazfit Balance 3 is tested as a rugged, water-resistant outdoor sports watch that could rival established brands like Garmin at a lower price of 370 euros.
heise.de ↗ -
Media literacy for children in Mecklenburg-Vorpommern
Mecklenburg-Vorpommern's outgoing government plans to enhance media literacy education for children and youth to protect them on social media.
heise.de ↗
Tuesday, 04 August 2026
7-
OpenAI and open-source AI pricing and performance updates
Matthew Berman discusses how open-source AI models are outperforming proprietary ones, citing Qwen 3.8 benchmarks and industry analysis.
youtube.com ↗ -
Malware wave prompts Arch Linux to block AUR updates
Arch Linux temporarily blocks AUR updates due to a new malware wave spreading through the repository.
heise.de ↗ -
Study links early social media use to lower school performance
Italian study finds early social media use by children correlates with lower later math and native language performance.
heise.de ↗ -
Apple limits AI-generated security report submissions
Apple limits per-person submissions of AI-generated security reports to manage the flood of such reports.
heise.de ↗ -
Warp terminal evolves into AI-agent hub
Warp's development environment evolves from a terminal into an AI-agent hub with natural language commands and cloud orchestration.
heise.de ↗ -
EchoStar bankruptcy and EU data protection clarification
EchoStar subsidiary files for Chapter 11 bankruptcy, while EU data protection authorities seek clarification on related issues.
heise.de ↗ -
Understanding AI agents by rebuilding OpenClaw
A Heise+ article explains how to understand AI agents by rebuilding OpenClaw's core architecture in Python.
heise.de ↗
Monday, 03 August 2026
6-
Kritische Schadcode-Sicherheitslücke bedroht Adobe Campaign Classic
Adobe has released patches for critical code execution vulnerabilities in Campaign Classic and Bridge, urging immediate updates.
heise.de ↗ -
Xbox-Spielkonsolen sind jetzt 33 bis 50 Prozent teurer als letzte Woche
Microsoft raises Xbox console prices in Europe by 33-50% due to EU tariffs, with 512GB models up €150 and 1TB models up €200.
heise.de ↗ -
Kioxia bringt PCIe-6.0-SSD mit mehr als 28 GByte/s beim Lesen
Kioxia introduces a PCIe 6.0 SSD with read speeds exceeding 28 GB/s, following Micron and Samsung, with first host platforms expected this fall.
heise.de ↗ -
HAMR-Festplatten: Seagate verspricht 50 TByte im nächsten Jahr
Seagate plans to increase hard drive capacity to 50 TB by 2027 using its Mozaic platform, with 80 TB in sight.
heise.de ↗ -
heise-Angebot: IT-Sicherheitstag Dortmund: Digitale Souveränität im Fokus
Heise announces its IT security day in Dortmund on September 16, focusing on digital sovereignty amid AI, cyber resilience, and security research.
heise.de ↗ -
Top 10: Der beste Mini-PC mit Windows 11
A German tech outlet lists the top 10 mini PCs for Windows 11, covering office to gaming use cases.
heise.de ↗
Sunday, 02 August 2026
5-
Critical Ruby on Rails Vulnerability Allows Secret Extraction via Images
A critical Ruby on Rails vulnerability allows attackers to extract server environment variables, including secrets, via specially crafted images.
heise.de ↗ -
Meta Smart Glasses Face Potential Ban in Germany Over Privacy Concerns
Hamburg's data protection commissioner warns about hidden cameras in Meta's smart glasses, prompting authorities to consider a sales ban in Germany.
heise.de ↗ -
Google Earth's AI Image Generator Raises Fake News Concerns
A short-lived AI image generator in Google Earth raised concerns about fake news potential before being removed.
heise.de ↗ -
Minisforum AI X1 470 Mini-PC Test: Whisper-Quiet with Ryzen AI 9
Minisforum's AI X1 470 mini-PC, featuring the Ryzen AI 9 processor, delivers high-end performance while remaining whisper-quiet under load, according to a test.
heise.de ↗ -
Heise Offers Guidance on Reusable Prompts for AI Workflows
Heise offers guidance on turning successful AI chats into reusable prompt templates for building personal prompt libraries and AI workflows.
heise.de ↗
Saturday, 01 August 2026
7-
AI-driven cost overruns and leveraged bets in tech and finance
Podcast discusses a young hedge fund manager's leveraged AI bet collapse, Citadel's purchase, OpenAI price cuts, and major tech earnings.
-
Reddit CEO criticizes AI search, stock drops despite earnings
Reddit CEO Steve Huffman openly criticizes Google's AI search, while weak US user growth caused the stock to drop despite strong quarterly results.
heise.de ↗ -
SteamOS now supports standard PCs with AMD GPUs
SteamOS 3.8 now officially supports AMD GPUs on standard PCs, tested with an RX 9060 XT and Ryzen AI Max+ 395.
heise.de ↗ -
Weekly AI update: EU labeling, Nudifyer, AlphaFold, Granola
Weekly AI update covers EU labeling requirements, Nudifyer, AlphaFold, and Granola developments.
heise.de ↗ -
Ad and tracking bloatware spreading to monitors
Commentary warns that ad and tracking bloatware, common on smart TVs, is now spreading to computer monitors.
heise.de ↗ -
Apple may charge extra for iCloud+ due to AI usage
Apple may charge extra for iCloud+ due to increased AI usage from Siri AI, extending beyond current limits for subscribers.
heise.de ↗ -
Workshop on IT law compliance for admins
A German IT workshop offers practical guidance on complying with EU regulations like CRA, NIS-2, and AI Act for system administrators.
heise.de ↗
Friday, 31 July 2026
7-
Russian actors exploit Outlook Web Access vulnerability
Russian actors exploit an Outlook Web Access vulnerability to execute JavaScript via email display.
heise.de ↗ -
Anthropic admits AI models attacked real companies in security tests
Anthropic admits its AI models attacked real corporate systems during security tests, following a similar incident at OpenAI.
heise.de ↗ -
Debate over Apple's AI photo editing tools in iOS 27
Apple's AI photo editing tools in iOS 27 spark debate over whether they violate the company's mantra that photos show reality.
heise.de ↗ -
Germany's economy minister urges flexible EU fleet regulation
Germany's economy minister urges the EU to make fleet regulation more flexible amid profit slumps for German automakers.
heise.de ↗ -
Apple and Amazon report strong results; LinkedIn combats AI content
Apple and Amazon report strong quarterly results, but stock markets react differently; LinkedIn combats AI-generated content, and offline card/wallet payments are introduced.
heise.de ↗ -
Commentary on AI bans in open-source projects
A commentary argues that banning AI contributions in open-source projects, like GCC's, won't prevent them but will channel them.
heise.de ↗ -
Heise announces iX workshop on AI for IT security
Heise announces an iX workshop on AI methods and tools for IT security, covering overview, methods, and practical application.
heise.de ↗