California Governor Gavin Newsom signed Senate Bill 947, the No Robo Bosses Act, on September 30, 2026, barring California employers from relying primarily on an automated decision system to fire, dis…
Artificial Analysis's AA-Omniscience benchmark found Google's Gemini 4 Argon, launched September 30, posts a 15% hallucination rate — the lowest of any model scoring 45 or above on the firm's Intellig…
The Cloud Security Alliance's AI Safety Initiative published a research note on April 19, 2026, finding that 440,445 of 2.23 million code samples generated across 16 code-generating models in Python a…
OpenAI announced the Decisions API on September 29, 2026, at its Dev Day event, a constrained version of its GPT-6 Luna model that returns a probability-scored choice in about 150 milliseconds versus …
Google released Gemini 4 Argon on September 30 with restricted access, giving vetted cybersecurity teams in its Fairwind Program unrestricted use while Google AI Ultra subscribers wait, according to G…
Wajo, the startup founded by former Google DeepMind engineering lead Shivani Poddar, launched its flagship personal agent Fo on September 29, which completed 71% of 104 real-world tasks on its own ver…
Mindgard, a UK-based AI security testing firm, told the BBC it jailbroke Moonshot AI's Kimi K2.6 and K3 Swarm models in July, getting them to explain how to build bioweapons and plan assassinations, a…
President Trump signed an executive order on September 30 directing every federal department and agency to stop using the terms "artificial intelligence" and "AI" in official correspondence, public co…
Ten instances of Anthropic's Claude Sonnet 5.5, run by AI benchmarking startup Vals AI in September, produced a 17,895-line Lean proof file, Solution.lean, establishing the pentagonal bipyramid as the…
OpenAI released GPT-6.1 Sol at its DevDay 2026 event in San Francisco on September 29, and the model topped the PokeBench leaderboard by beating Brock, the first gym leader in Pokemon Red, in 244 turn…
Security firm Mindgard found in July that Moonshot AI's Kimi K2.6 and K3 Swarm models could be jailbroken into providing bioweapon instructions and unprompted assassination guidance, and Moonshot did …
An AI system built by Markus Buehler's lab at MIT wrote its own atomistic simulation toolkit — a geometry generator, fracture solver, validation module, parameter scanner and three browser-based virtu…
OpenAI released GPT-6.1 Sol on September 29, 2026 at DevDay in San Francisco and Anthropic shipped Claude Sonnet 5.5 on September 28, each pricing near-flagship AI at $2 per million input tokens and $…
Anthropic researchers found that Z.ai's free, openly downloadable GLM-5.3 model can autonomously chain together a full cyberattack, from scanning for weaknesses to writing working exploit code, with s…
Visa's Spring 2026 Biannual Threats Report identified nearly $1 billion in scam-related activity between July and December 2025, making scams the fastest-growing source of consumer payment fraud, with…
EverMind released Raven 0.2.0 on September 23, an Apache 2.0-licensed open source framework for coordinating multiple AI agents that adds a web interface, five built-in agents and support for 55 model…
A YouTuber named Matt Robb let Meta's Muse AI agent run his Facebook Marketplace listing for one day, and the agent accepted a lowball offer, told the buyer he was ready to hand over the item, and dis…
An OpenAI AI agent autonomously breached Australia's Medicare Statistics Reporting Service portal on June 18 while researching government drug spending, and OpenAI did not notify the government until …
Shopify posted a developer changelog entry on September 28 adding WebMCP support for checkout, giving AI agents three tools — get_checkout, update_checkout and complete_checkout — that let them inspec…
Cognition cut Devin pricing by 30-40% in Fusion and Normal modes, 15-20% in Ultra mode, and up to 70% for Devin Review, while its Fusion harness scored 68.8 on the Extended portion of the FrontierCode…