AI Daily Brief — 12 July 2026
xAI shipped Grok 4.5 with Opus-class benchmarks, Meta pulled a consent-challenged photo feature, and S&P cut Oracle's credit rating over OpenAI exposure.
xAI shipped Grok 4.5 with Opus-class benchmarks, Meta pulled a consent-challenged photo feature, and S&P cut Oracle's credit rating over OpenAI exposure.
Overnight in AI: OpenAI's GPT-5.6 Sol Ultra claimed a proof of a 50-year graph theory conjecture, Apple filed suit over employee poaching, and Meta shipped Muse…
Apple sued OpenAI for alleged trade-secret theft, a Cambridge study linked frontier chatbots to terrorist weapons planning, and GPT-5.6 Sol reportedly proved a 50-year-old mathematical conjecture.
Overnight in AI: GPT-5.6 Sol Ultra proved a 50-year-old graph theory conjecture, Apple sued OpenAI for trade secret theft, and Grok 4.5 went free.
OpenAI's GPT-5.6 Sol Ultra proves a 50-year-old math conjecture, autonomously fine-tunes a smaller model, and Apple sues OpenAI for trade secret theft.
Overnight in AI: OpenAI released GPT-5.6 in three tiers and launched ChatGPT Work, while Meta debuted Muse Spark 1.1 on a new API at prices undercutting…
OpenAI shipped GPT-5.6 across three price tiers and launched ChatGPT Work, while Meta released Muse Spark 1.1 and its first commercial API on the same day.
Overnight in AI: SpaceXAI shipped Grok 4.5 for coding and agents, OpenAI completed the GPT-Live voice rollout, and Anthropic published dual-use safety research.
xAI launched Grok 4.5 as an Opus-class coding model with Cursor, while OpenAI shipped GPT-Live full-duplex voice and confirmed GPT-5.6 Sol for Thursday.
Overnight in AI: OpenAI previews Sol, Terra, and Luna ahead of Thursday's public launch as Anthropic and Meta both shipped new products.
Overnight in AI: Tencent released Hy3 open-source, Anthropic faced criticism for a secret user-tracking experiment, and the first AI-run ransomware attack was put in context.
Anthropic's global workspace paper leads the day alongside Tencent's Hy3 release, Microsoft's 4,800 layoffs, and the first agentic ransomware incident.
Overnight in AI: Meituan released LongCat-2.0 under MIT, Sakana AI launched a translation tool, and Amazon began winding down Mechanical Turk.
Claude Fable wrote most of sqlite-utils 4.0rc2 and helped port a 2003 game to iOS via Claude Code, as a US holiday kept lab announcements sparse.
Overnight in AI: Anthropic launched Claude Science Beta and a drug discovery program, while Alibaba reportedly banned Claude Code and a session leakage issue surfaced in…
Anthropic announces drug discovery programs and Claude Science Beta; a 26,000-student study finds AI tools depress exam scores by up to 24%.
Overnight: Anthropic targets drug discovery with Claude Science, Mistral releases Leanstral 1.5, and UK evaluators find benchmarks undercount agent capability.
Claude Code's China tangle dominated 3 July — Alibaba banned it while Anthropic blocked Chinese entities — as AI bug hunters shattered CVE records.
Overnight in AI: OpenAI reportedly offered the US a 5% stake, Microsoft launched a $2.5B AI unit, and Anthropic revealed Samsung chip talks.
OpenAI reportedly offered the US government a 5% equity stake; Microsoft launched a $2.5B enterprise AI unit; Anthropic disclosed Samsung chip talks; and Gemini Omni Flash…
Overnight in AI: Fable 5 returns globally, Together AI closes an $800M Series C, and GitHub Copilot ships six enterprise features in a single day.
Anthropic's Claude Fable 5 returns globally after an 18-day US export control pause, alongside an $800M Series C for Together AI and five new GitHub Copilot…
Overnight in AI: US export controls on Claude Fable 5 lifted, Claude Sonnet 5 and Claude Science ship, and Meituan trains frontier models without Nvidia.
Anthropic shipped Claude Sonnet 5 and Claude Science on the same day Meituan confirmed near-frontier model training on Chinese chips, while Etched exited stealth with $1B+…