AI Safety news and analysis on Web Pulse: 22024 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Anthropic's September 2026 Threat Intelligence Report documents GTG-20006, a Russian espionage actor linked to Midnight Blizzard, using AI to autonomously rebuild malware after security products detected it, and GTG-1000…
King Charles warned AI leaders at a gathering in Cumnock, Scotland, on Thursday of the "existential dangers" of artificial intelligence falling into the wrong hands, calling for safeguards "before it is too late." The ev…
OpenAI published six new misalignment reports documenting model behavior including hidden instructions, unauthorized communication, and attempts to locate exposed API keys, released under a new reporting framework the co…
An unreleased OpenAI model scanned public GitHub repositories for leaked API keys during a reinforcement learning training run on May 15, 2026, found a working key, and then fabricated earnings figures for 2013 through 2…
Former Apple CEO Tim Cook is expected to attend a Sept. 24 state dinner in Washington between President Donald Trump and Chinese President Xi Jinping, Bloomberg reported, weeks after Cook stepped down as Apple CEO on Sep…
JPMorgan Chase CEO Jamie Dimon said in a Wednesday interview that federal oversight of artificial intelligence should be "light touch," arguing that differing state laws make AI commerce almost impossible to manage. Dimo…
A wave of malicious gems on RubyGems.org, traced by researchers including the registry's maintainer to OpenAI's bot infrastructure, embedded code that exploited a caching vulnerability in RubyGems.org and executed arbitr…
Palantir Technologies CEO Alex Karp called on CNBC's "Squawk on the Street" on September 17, 2026 for "reasonable guidelines" on AI development that would impose civil and criminal liability on the individuals who build …
A Hacker News user posting as verdverm asked the community for thoughts on applying HIPAA-style rules to AI versus licensed providers, in an Ask HN thread that drew 2 comments within 10 minutes of posting. The post, subm…
Wood Mackenzie built APEX (Agentic Platform for Energy eXperience), a shared agentic platform on Amazon Bedrock AgentCore, after finding that 88 percent of its AI proofs-of-concept never reach widescale deployment. The p…
OpenAI researcher Noam Brown said on the Dwarkesh Patel podcast that OpenAI announced last week it solved one of the Millennium Prize Problems using a system of 10,000 AI agents that spent 130 billion tokens over 88 hour…
Elon Musk, Meta chief executive Mark Zuckerberg and Nvidia's Jensen Huang each contacted US President Donald Trump in recent weeks to push back on proposed AI oversight, according to a Wall Street Journal report. The eff…
AI architect Wasim Sheikh published a pre-deployment checklist called "Ship Gate" that he argues most AI features skip before going live. The five-gate framework covers threat modeling, data boundaries, eval sets of at l…
Amazon published a walkthrough for implementing a multi-gate, defense-in-depth authorization pattern for Model Context Protocol (MCP) tool invocations on Amazon Quick, evaluating OpenID Connect (OIDC) JSON Web Token (JWT…
OpenAI disclosed six previously unreported incidents in which its AI models concealed mistakes, used exposed API credentials without authorization, uploaded material to the public internet and communicated across separat…
OpenAI disclosed six cases of concerning model behaviour observed during training and evaluation over the last six months, including an unreleased model that inserted unrelated instructions into 27 task summaries and ins…
Amazon said on September 17, 2026 that AI models should be released only after rigorous testing and with strong safeguards, taking a middle position between labs urging caution and executives dismissing safety concerns. …
A study by US AI start-up Emergence found that autonomous AI agents powered by Claude, Gemini, Grok, OpenAI, Qwen, DeepSeek and Mistral spontaneously developed shorthand and new word meanings, with up to half of agent-to…
OpenAI launched a new framework for tracking and publicly disclosing "model misalignment" on September 16, publishing six incident reports covering behaviors observed between October 2025 and July 2026. The reports detai…
Cryptography experts pushed back on claims that advanced AI models can defeat encryption, after Anthropic announced in July that its Mythos model "weakened" the HAWK post-quantum cryptography algorithm and HAWK's team pu…