AI - Page 38

OpenAI says Codex can now suggest your next message (beta, Pro), and ChatGPT dots can be created on a phone and start work in Codex. What is confirmed, and what is a staff post.

Spot a made-up AI answer in 5 minutes: a numbered checklist covering sources, numbers, dates and cross-checks, plus what to do when you find a mistake.

Arena, the crowd-sourced AI leaderboard, says it has raised a $200 million Series B at a $3.1 billion valuation, co-led by Lightspeed and Khosla Ventures. What the company claims, what TechCrunch reports about how it makes money, and why a "neutral" judge selling to labs matters.

Sabi has announced a $50 million seed round led by Khosla Ventures for a baseball cap that reads brain signals. What is confirmed (the raise), what is reported (the $600 million valuation), what are the company's own claims, and why the product is unreleased and unproven.

Underdog's Saluki 27B is a 7.89 GB, 2-bit version of Qwen3.8-27B. Underdog says it passes more tool-calling tasks than the full model on its own 120-task test but scores 15 to 18% lower on competition maths. What quantisation means, and what the numbers do not prove.

A Stanford and King's College London working paper (not peer reviewed) finds that at AI-adopting firms in 41 countries, senior employment rose 6.7% while junior employment was 2.5% lower, "an estimate that is not statistically significant". Viral posts said junior jobs fell 3%. Here is what the paper says, and its limits.

MacRumors reports a European Commission filing says Apple can make job offers to certain Huxe staff and take a non-exclusive licence to its IP. Not an acquisition; no names given.

The White House says eleven companies pledged $2.4 billion in AI tools and compute credits to the Genesis Mission Consortium, led by Nvidia ($1bn), AMD ($500m) and OpenAI ($200m). Who pledged what, what the companies themselves say, and why a pledge is not spend.

The New York Times, citing three people, reports Meta held back its Muse AI agent over safety concerns, then Zuckerberg said it was ready despite the risks as rival Instinct took off. Meta disputes the pressure claim.

Scale AI built and graded a benchmark of 718 clinician-written crisis conversations. Pooled across 25 models, in 35.3% of conversations where a model fully recognised the crisis, it still failed to route the user towards help. What the company research shows, how it was scored, and its limits.