AI News for July 13, 2026 — 7min.ai

[NEWS] S&P downgrades Oracle, names OpenAI a 'key credit risk': S&P Global cut Oracle's rating from BBB to BBB-, one notch above junk, and flagged OpenAI as a key credit risk.
[RESEARCH] MIT method detects CSAM-generating models without producing any images: MIT researchers and child-safety nonprofit Thorn built an auditing technique that determines whether an open-source model can generate child sexual abuse material without ever prompting it to produce output.
[NEWS] OpenAI names new VP of research and safety after reorganization: OpenAI created a new vice president role for research and safety, merging its previously separate Safety Systems team into the broader research organization.
[NEWS] SK Hynix shares slump 13% after $26.5B Nasdaq debut euphoria fades: SK Hynix's American depositary receipts fell as much as 13.2% in Seoul trading on Monday, just one session after their blockbuster $26.5 billion Nasdaq debut.
[TOOLS] Anthropic adds a built-in browser to Claude Code: Anthropic gave Claude Code an integrated, tab-based browser window that lets the model open, read, click, and type on external web pages directly inside the app, including documentation sites and issue trackers.
[RESEARCH] Anthropic: Claude Cowork's top job is the office work nobody wants: An Anthropic analysis of 1.2 million anonymized Claude Cowork sessions from May found that about half of all usage falls into two buckets: business operations and text-based work.
[NEWS] Altman now 'pretty sure' AI is net job-creating, reversing earlier warnings: OpenAI CEO Sam Altman wrote that he's 'pretty sure' AI has been net job-creating so far, adding 'this is not what I expected.' The stance reverses his earlier warning that AI's labor impact could arrive fast enough to be 'a little scary.' Anthropic CEO Dario Amodei has similarly walked back predictions, now framing automation as a productivity multiplier rather than a job killer after once forecasting 20% unemployment.
[RESEARCH] Class average fell from 96% to 48% when a Brown professor banned AI on exams: Brown economics professor Roberto Serrano suspected most of his 86 students used AI to cheat after a take-home exam averaged 96%, far above the usual 65-80%.
[RESEARCH] Structured memory lets AI agents finally win at Slay the Spire 2: Researchers at Alaya Lab, Shanghai Jiao Tong University, and others built AgenticSTS, an agent architecture that never sees its own chat log and instead rebuilds each decision from five organized memory layers.
[RESEARCH] AI lifts individual research careers but flattens science overall, Nature study finds: An analysis of more than 40 million academic papers found that scientists using AI tools publish more, collect more citations, and reach leadership roles sooner than peers who don't.

About 7min.ai · Subscribe to newsletter