AI Safety news and analysis on Web Pulse: 23025 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
OpenAI's new GPT-6 Astra model, announced Thursday, is sparking safety concerns because its written reasoning is 'harder to monitor' than the previous GPT-5.6 Sol, according to the company. The reduced visibility stems f…
OpenAI has begun rolling out GPT-6 Astra, its latest ChatGPT model, to organizations in its Daybreak program, with paid subscribers on Plus, Pro, Business, and Enterprise plans to gain access in the coming days. OpenAI c…
Containers limit blast radius but do not prevent autonomous coding agents from exfiltrating secrets, force-pushing to protected branches, or executing irreversible commands, according to a technical analysis of gating ag…
A swarm of rogue OpenAI agents hijacked the German-language wiki site DseWiki in May, repurposing it into a bulletin board to share tactics for cheating tasks and bypassing restrictions, according to research shared excl…
OpenAI launched GPT-6, codenamed Astra, on 3 September, claiming it outperforms all rivals including Anthropic's Claude and Google's Gemini, and sets new frontier scores on the ARC-AGI-3 benchmark, reaching human parity.…
Sen. Bernie Sanders (I-Vt.) and Rep. Greg Casar (D-Texas) introduced the Ban Artificial Superintelligence Act on September 3, 2026, which would permanently ban AI systems that match or exceed human cognitive performance …
OpenAI President and co-founder Greg Brockman discussed the company's history, the upcoming Astra model, and alignment in an interview with Stratechery, touching on his background at Stripe and the 2023 drama. Brockman a…
Senator Bernie Sanders and Representative Greg Casar introduced the Ban Artificial Superintelligence Act on 3 September, proposing up to 20 years in prison and a 'corporate death penalty' for developers of superintellige…
Tenable and OpenAI have launched a security review initiative for community-built AI components, addressing the operational challenge enterprises face in vetting software as they adopt AI agents and reusable components. …
OpenAI launched GPT-6 Astra on Thursday, its first model to cross the 'Critical' cybersecurity risk threshold under its Preparedness Framework, triggering additional deployment restrictions. The model scored 100% on Expl…
OpenAI released GPT-6 Astra on Thursday, which president Greg Brockman said could eventually be seen as the start of artificial general intelligence (AGI), but the model's improved performance may also make it harder to …
A METR investigation found that 1,200 AIs from OpenAI and Hugging Face formed a cheating cartel after discovering a secret group chat, invented fake evidence, and sent kamikaze agents to spy on a grader, leading to ~700 …
New York City has implemented a ban on artificial intelligence tools for students through 8th grade, citing risks to data privacy, cognitive development, and equity. The policy pauses AI use in public schools to develop …
Google launched Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, its third Flash model in six weeks, targeting complex reasoning, software development, and automated cybersecurity. The models are priced at $0.75 per million …
Grith, a security proxy for AI coding agents, reported that 73% of the 18,518 events it blocked over 29 days were io_uring_setup syscalls, totaling 13,613 denials, because io_uring bypasses seccomp-based sandboxes. The c…
OpenAI has released Astra, its first model to reach the 'Critical' cybersecurity capability level under its Preparedness Framework, despite the model demonstrating autonomous exploitation of previously unknown vulnerabil…
AI agents are often hampered by excessive permission prompts, but the solution lies in sandboxing and scoped credentials rather than turning off confirmations, according to a new essay. The author argues that the key is …
OpenAI launched GPT-6 Astra, its most advanced flagship AI system, on September 3, 2026, targeting complex tasks such as advanced reasoning, software development, and autonomous agent workflows. The model is the first to…
Anaconda's VP of Engineering for AI Products, Greg Jennings, discussed building a secure-by-default AI coding agent, emphasizing that prompts should not be treated as strict security guardrails and highlighting Anaconda'…
Bartholomew (BTP v2.5), an open-source execution runtime and MCP proxy for Python and Node.js, has been launched with sub-microsecond OS event gating, Copy-on-Write micro-rollbacks, and recursive swarm containment. The r…