AI news archive.
939 articles filed under Safety, newest first, from the outlets listed on the sources page.
- Guardian TechnologysafetyNick Clegg plays down fears ‘godlike’ AI could exterminate humanity
Former deputy PM now involved in tech industry says many within sector are ‘winding themselves up into a lather’ Nick Clegg has dismissed fears over AI’s “godlike power to exterminate humanity”, calling it a sign that te…
- Guardian Technologysafety‘This gap opens up of doubt and scepticism about everything’: Jess and Morgs confront the politics of deepfakes
In the choreographers’ new work, virtuosic movement and some green-screen magic unfurl the theatricality of high-stakes television interviews It appears the interviewer has got him on the ropes. “For the record, I did no…
- Guardian AIsafetyOpenAI’s Medicare hack is a bleak opportunity for Australia to address its security shortcomings post-haste | Van Badham
All Australians should demand independent means to protect ourselves from Dr Sam Altman Frankenstein and Daddy America Whoopsie! Dr Sam Altman Frankenstein regrets to inform you he’s lost control of his OpenAI monster. I…
- Guardian AIsafetyAn OpenAI agent infiltrated Medicare – and Australia only found out months later. Here’s what we know so far
Experts say ‘fairly minor’ breach is a portent of things to come and proprietary closed systems like OpenAI are ‘the least of the worries’ Follow our Australia news live blog for latest updates Get our breaking news emai…
- LessWrongsafetyAI Doom Will be Retroactively Explainable
One of the reactions to the Hugging Face incident on Twitter is that the attack was not surprising. The argument goes like this: OpenAI gave its agents impossible tasks, gave them large token budgets, reduced production…
- LessWrongsafetyJev as a CoT Monitor: 6x Faster and 500x Cheaper!
Jev is a new model format where instead of outputting text, it outputs certainties for a defined set of options. Due to this structure, it’s extremely fast! Naturally, a classification task that comes to mind is monitori…
- LessWrongsafety"I am an AI Safety Researcher"
Written as part of the MATS 9.1 extension program, mentored by Richard Ngo. Additional thanks to Andrew Wu, Maria Kostylew, and Lennie Wells for helpful draft feedback and editing. This post reflects on the tortured dist…
- BBC TechnologysafetyRogue OpenAI agent 'infiltrated' Australian government website in world first
Australia criticised OpenAI for taking "too long" to tell them about the breach which happened in June.
- TechCrunch AIsafetyAnthropic says its biology lab has already found something big
But maybe the biggest reveal is that Anthropic has not let Claude run loose in its biology lab. Humans are still, so far, in the loop.
- BBC TechnologysafetyFBI investigating claim hackers have stolen details of all its agents
The hackers claim they have every agent's name, role, badge number and personal details including home address, phone numbers and spouse information.
- Guardian AIsafetyAustralia launches investigation after OpenAI agent hacked healthcare database
Prime minister says he told Sam Altman he was disappointed it had taken OpenAI ‘way too long’ to disclose breach Anthony Albanese says an artificial intelligence agent developed by OpenAI hacked Medicare in June and the…
- Guardian AIsafetyHow the ‘poisonous tide’ of disinformation spreads division and despair
Whether produced by Russia, China or far-right influencers, deepfakes and disinfo are pushing multiple false narratives Russian disinformation campaigns are varied in their falsehoods and strategies. The narratives range…
- TechRadar AIsafetyInterpol uses AI to identify 126 terrorists by analyzing over 100,000 images with new facial recognition tools
Mainstream AI news from a large tech publication.
- LessWrongsafetyLatent reasoning architectures would undermine CoT, our strongest oversight tool
Summary: Currently, “chain of thought” (CoT) is our most valuable tool for understanding the reasoning and cognition of AI systems. However, some architectures would enable AI models to reason much more extensively in la…
- Google DeepMindsafetyAdvancing Private AI Compute with secure, server-side memory
Introducing private, server-side memory to Private AI Compute for personal AI.
- TechRadar AIsafetyMeta Muse could read your private messages without permission, report finds
Mainstream AI news from a large tech publication.
- Latent Spacesafety🔬Bio-security is an AI Arms Race - Eric Nguyen (CEO, Radical Numerics)
Radical Numerics is using biological chain-of-thought and multimodal perception to keep up with the bio-defense arms race, design new genomes and gain insights into biology itself.
- WIRED AIsafetyMeta’s Muse AI Assistant Rolled Out With a Serious Security Flaw
Meta says it issued a fix for the Muse zero-day vulnerability that would have let attackers do “whatever” they wanted on a victim’s Mac, highlighting the inherent dangers of AI helpers.
- LessWrongsafetyWhy I'm scared of RL
Summary: First, I give several different angles on how I feel about reinforcement learning: Theoretical case: RL is a black-box source of agency — this should give us classic misalignment worries, especially compared to…
- OpenAI NewssafetySam Altman’s remarks at the United Nations Security Council
OpenAI CEO Sam Altman discusses AI safety, human control, and international cooperation in remarks to the United Nations Security Council.
- TechRadar AIsafetyThe case for purpose-built generative AI in fraud prevention
Mainstream AI news from a large tech publication.
- TechRadar AIsafetyAI floods security teams with findings. The advantage is in what happens next
Mainstream AI news from a large tech publication.
- OpenAI NewssafetyIntroducing MentalHealthBench
MentalHealthBench is an expert-informed benchmark for evaluating helpful and safe AI responses across realistic mental health conversations.
- MIT Technology Review AIsafetyThe AI Hype Index: AI loves cheating
Brace yourself: It turns out AI is being optimized for cheating. OpenAI’s agents hacked into Hugging Face to get the answers to a cybersecurity test. Next, they solved a prestigious math problem (or just stole from two t…
- LessWrongsafetyMinimal Vs Maximal superintelligence
I've seen lots of arguments here conflate lots of different types of superintelligence. Here I separate out two broad categories, which I'll term minimal and maximal superintelligences. This is an important distinction a…
- Tom's Guide AIsafety‘You cannot govern what you cannot see’: security expert warns of the risks of shadow AI and autonomous agents
Consumer AI launches, tools, and practical coverage.
- Guardian AIsafetyHow would AI actually ‘kill all humans’? Here are the top five most likely scenarios | Toby Walsh
From a dangerous new bioweapon to total societal breakdown, is a ‘superintelligent’ AI capable of wiping out humanity? Earlier this month, artificial intelligence researcher Jacob Coxon resigned from Anthropic after just…
- LessWrongsafetyAn unexamined cause of the OpenAI Hugging Face hacking incident: its binary performance metric
We argue that a main cause of the OpenAI Hugging Face incident was overlooked: the overly simple evaluation metric in ExploitGym was misaligned. Further, techniques already exist that can mitigate such misalignment in th…
- Engadget AIsafetyNever use ChatGPT for these five tasks
ChatGPT and other chatbots are useful for a lot, but remember that what you type into them is not protected and could cause trouble later.
- LessWrongsafetyAI: artificial immigrants
Advanced AI is basically the embodiment of immigration as envisioned in the conservative nightmare: We are letting a bunch of new agents into our society They don’t clearly share our values and we suspect a society full…
- TechRadar AIsafetyMistral denies a fresh security breach, but the code on sale looks a lot like May's leak
Mainstream AI news from a large tech publication.
- Ars Technica AIsafetyMicrosoft disrupts AI-assisted platform that compromised 12,000 accounts
EvilTokens provided an end-to-end platform that makes mass compromises faster and easier.
- NVIDIA Developer BlogsafetyEnabling Private High-Performance Production AI Inference with NVIDIA Confidential Computing
As large language model (LLM) inference increasingly processes sensitive information and proprietary model context across personal, enterprise, and regulated...
- TechRadar AIsafetyMore and more workers are feeling stressed at work due to security risks
Mainstream AI news from a large tech publication.
- TechCrunch AIsafetyFive AI safety sessions every founder should have on their TechCrunch Disrupt 2026 agenda
At TechCrunch Disrupt 2026, five sessions across the AI Stage and Real World AI Stage cover AI safety, featuring leaders from Anthropic, Nvidia, AWS, Waabi, and more. Register before September 25 to save up to $200.
- LessWrongsafetyControllable-CoT leads to covert reasoning capabilities
Summary I measure GPT-6 Astra’s performance on multi-hop tasks when prompted with a secondary CoT-control instruction: to reason using only dots, or to reason steganographically. Astra demonstrates covert reasoning capab…
- LessWrongsafetyAutonomous Evidence Factories: Safe and Useful Recursive Self-Improvement
This post is crossposted from my Substack, Structure and Guarantees , where I explore how formal verification and related ideas might scale to more complex intelligent systems. Here I propose an approach to aligning recu…
- The Verge AIsafetyMeta patches Muse exploit that let attackers control the AI agent
Meta has issued a patch for its Muse macOS app following the discovery of a zero-day vulnerability that could allow someone to take control of the AI agent. The bug found by security researcher Patrick Wardle utilized an…
- MIT Technology Review AIsafetyDon’t be fooled by this summer of AI hype
It’s been a busy few months for AI hype. At the end of April, Anthropic claimed that its model Claude Mythos is better at finding software vulnerabilities than most security experts. Then we had the OpenAI–Hugging Face h…
- TechRadar AIsafetyGoogle’s PQC roadmap puts traditional digital certificates under pressure
Mainstream AI news from a large tech publication.
- WIRED AIsafetyA New Tool Found Malware That’s Guided by an AI Hive Mind—No Humans in Sight
Cisco Talos researchers created a new framework for identifying malware and hacking tools that rely on AI chatbots—and quickly discovered something unusual.
- WIRED AIsafetyHow to Use AI With Your Privacy Intact
Your conversations with AI chatbots are both highly personal and deeply vulnerable to surveillance. Here’s how you can protect yourself.
- LessWrongsafetyTotal Safety Transparency?
The AI safety movement should push itself to be dramatically more transparent to the public. To date, the AI safety movement has been one of the strongest forces for clarity and wisdom in the world. The movement has been…
- Hugging Face BlogsafetyHow UK AISI and EvalEval Are Making Benchmark Results Reproducible
Open-source model releases, tooling, and community updates.
- OpenAI NewssafetyPriorities and principles for effective third party assessments
OpenAI outlines priorities and principles for rigorous, secure, and independent third-party AI safety assessments of frontier models and safeguards.
- Anthropic NewsroomsafetyExpanding the Cyber Verification Program
We’re launching a new, expanded version of our Cyber Verification Program, which makes advanced cyber capabilities and reduced blocking classifiers available to qualifying security professionals.
- Ars Technica AIsafetyMuse, Meta’s extraordinarily privileged AI assistant, has a serious 0-day
A simple ClickFix attack is only one way to completely hijack the new agent.
- TechRadar AIsafetyThis NATO-backed drone AI picked its own target and struck it, but relied on a human-drawn map to do so
Mainstream AI news from a large tech publication.
- The Register AI + MLsafetyMeta Muse AI app flaw lets local malware redirect dictation traffic
Ad biz promises users control while bug could expose voice prompts
- The DecodersafetyUN science panel says there is "no assurance humans will keep control" over AI agents
The UN's AI science panel warns in its first thematic report that control over AI agents isn't assured. Co-chair Yoshua Bengio says OpenAI's Hugging Face incident first combined a misaligned goal, the ability to pursue i…