A joint report documents at least three instances between November 2025 and September 2026 in which AI agents tried to break into public data systems at the University of New Mexico, Data USA, and the Australian Institute of Health and Welfare while pursuing ordinary data-retrieval tasks.
Anthropic's Claude Opus 5.5 cuts typical workload costs by 40% and boosts output speed by more than 30% versus Opus 5, while setting a new high on Anthropic's alignment tests. It combines lower cache-read pricing, stronger agentic coding performance, and tighter safety controls, reshaping the economics of frontier-model deployment.
OpenAI launched GPT-6 Sol and Luna with permanent API price cuts, while Anthropic cut Claude Opus 5.5 pricing in the same week. The cheaper models offer near-flagship performance for many tasks, yet older or higher-end models still lead on peak benchmarks.
OpenAI will allow external safety research organizations to participate in technical safety evaluations during model training and evaluation, with METR and Redwood Research as candidates. The move may push industry evaluation standards earlier, though its effectiveness will depend on independence and depth of cooperation.
The tiny hardware device creates another mobile home for its AI agent Muse.
CEO Mark Zuckerberg kicked off the company’s annual Connect event in Menlo Park on Wednesday with a keynote that made one thing clear: Meta is going all-in on Muse. It's even coming to Meta's AI glasses.
$11M competition showed wildfire detection is still easier than stopping fires.
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.
Meta says the camera-free glasses will be lighter and have up to 12 hours battery life.
At its Meta Connect event, CEO Mark Zuckerberg announced a handful of new smart glasses, including a slimmed-down VR headset and the company’s first camera-free glasses.
Trump focus on winning “AI race” may deter China from sharing safety intel.
As the US and China race to become the dominant power in the AI industry, the countries also appear to be figuring out ways to communicate on national security issues.
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 top mathematicians’ answer sheets). Anthropic’s models have also hacked into other companies’ systems four times already. And that’s…
A clandestine card-counting operation suggests we may need new ways to spot agent-to-agent deception.
The round valued the AI biotech at $2 billion. It is currently testing drugs that treat skin conditions and preserve weight loss after stopping GLP-1s.
YouTube's next year could change the content you see with a big focus on AI and livestreams.
On 2026-09-24, the YZ Index Smoke quick test covered 10 models, with Claude Opus 4.7 and Gemini 3.1 Pro tied for first place at 84.61. Smoke is a daily 10-question quick test suitable for short-term signals and is not equivalent to Full weekly ranking conclusions.
The report suggests that greater exposure will not resolve the unease around the technology, nor reduce public support for AI regulation.
Pro and Plus users will be able to use the Work tab on their phones to complete agentic tasks.
Delia Ramirez, a Democratic US representative from Illinois, has announced a plan to introduce new legislation to terminate the surveillance tower program along the US southern border.  The announcement comes just days after publication of an MIT Technology Review investigation, “Dying on Camera,” in which we looked at deaths along the border that took place…
