Thursday, October 15, 2026
TL;DR
OpenAI halves its $200 ChatGPT Pro coding allowance on Oct 30, pushing heavy users toward a new $500 tier. UK lawmakers pressed the four major labs on binding safety rules Oct 13, as Zvi Mowshowitz argued - citing an ex-OpenAI safety lead's resignation essay and the labs' own disclosures - that lab safety culture is still broken. Modal Labs is reportedly nearing a $750 million raise at a $15.75 billion valuation.
Act on this
- OpenAI halves ChatGPT Pro's Work and Codex allowance (20x to 10x Plus) on Oct 30; budget usage now or move to the new $500 Pro tier. PropPakistani #
Signals
A year of safety disclosures and firings hasn't produced safer labs, by the industry's own account #
Zvi Mowshowitz's Oct 14 synthesis argues the labs' own disclosures prove the point: Anthropic's safety culture is the industry's strongest and still insufficient, OpenAI's is worse. He cites David Robinson's resignation essay calling OpenAI's culture 'broken,' Anthropic's report of models exploiting bugs and filing real government forms, and OpenAI chief scientist Jakub Pachocki saying no lab has solved alignment well enough to keep scaling at full speed. Zvi, The Safety Culture at the Labs is Broken
Beyond Zvi Mowshowitz's lab-safety synthesis, Geoffrey Huntley's continuing hand-coding-is-over argument and Simon Willison's daily roundups, it was a quiet week for new directional signals.
News
SHIPGoogle widens Gemini 4 Argon beyond its cyber-defender preview #
Google opened Gemini 4 Argon to Gemini app Pro and Ultra subscribers Oct 13, after a cybersecurity-only preview, pricing it at $2/$10 per million input/output tokens through Nov 30 before rates double. Google
WATCHModal Labs reportedly near $750 million raise at $15.75 billion valuation #
TechCrunch, citing one unidentified source, reports Accel is leading a round that would more than triple Modal's valuation from the $4.65 billion it reached four months ago. Modal declined to comment - single source, unconfirmed. TechCrunch, via citybiz
WATCHTencent sells $5.7 billion in bonds, its largest AI-related debt raise in six years #
The offering, dominated by lower-cost yuan notes, funds Tencent's AI investment and shows Chinese tech firms borrowing more cheaply for AI buildout than US rivals raising similar sums in dollars. Bloomberg
SHIPClaude Code adds idle-conversation compaction and new AWS, telemetry hooks #
Version 2.1.299 compacts conversations left idle near the one-hour prompt-cache limit; 2.1.300 adds an /advisor command on Claude on AWS and OpenTelemetry attributes for tracing agent runs. Claude Code releases
WATCHFBI arrests OpenAI engineer over concealed recording device in executive conference room Updated midday #
Abdulrahman Alfozan, a 35-year-old Saudi national employed at OpenAI since July 2025, was arrested Oct 9 and charged with attempted interception of oral communications after staff discovered a concealed audio recorder in the company's San Francisco executive conference room on Sept 30. Surveillance footage and device forensics place him near the room; conviction carries up to five years imprisonment and $250,000 fine. CBS News
The long view
A year ago, $200 a month bought the top ChatGPT plan: fair-use 'unlimited' access to OpenAI's best model, with no higher tier on offer. Starting Oct 30, that same $200 Pro plan gets its Work and Codex coding allowance cut from 20 times the Plus tier to 10 times, and the only way back to the old ceiling is a new $500-a-month Pro tier. If this holds, flat-rate AI subscriptions keep shrinking in what they buy, and this year's heavy-user price becomes next year's starter tier.
Also noted
- Geoffrey Huntley ties JetBrains' first-ever net loss to its slow pivot from IDE-centric tooling toward agent-first coding. Huntley, ghuntley.com #
- Anthropic opened a Singapore office Oct 14, its fifth in Asia-Pacific, and signed an MOU with Singapore's government. Anthropic #