The Verge reports that the US government is approving sales of Nvidia's AI chips to Chinese firms with ties to blacklisted companies, including one buyer whose listed office was empty. Watcher Guru's relay of the report drew far more views. This is separate from the reporting on a Chinese spy plot targeting Jensen Huang. The scan did not find a response from Nvidia or the Commerce Department.

Watch for a response from the Commerce Department and Nvidia, and whether Congress adds this to its questions for Jensen Huang.

AI news account kimmonismus says a US judge allowed claims over Anthropic's Claude Max usage promises to proceed. Subscribers allege the advertised 5x and 20x usage multipliers applied per session while weekly allowances were much lower. Another post citing Courthouse News says other claims were dismissed. The ruling and the original article were not checked, so treat the details as unconfirmed.

Watch for the court ruling itself, a response from Anthropic and any change to how Claude plans describe usage limits.

The Information reports that Palo Alto Networks is in talks to pay $500 million for Clutch Security, an AI agent security startup that has raised more than $20 million. No deal has been announced.

Watch for a signed deal or confirmation from either company.

The Information reports that TSMC's push into memory is led by Hsu Kuo-chin, a vice president who oversees advanced packaging and previously ran Micron's Taiwan operations. The outlet says TSMC is weighing making advanced memory chips and that the effort could complicate its 2024 partnership with SK Hynix on high bandwidth memory for AI processors. No memory product has been announced.

Watch for TSMC confirming the effort and for SK Hynix's response.

Polymarket relays that Airbnb plans to launch its own AI travel agent in the first half of 2027, offering personalized trip planning. The underlying announcement appears to predate Nov 7, and this is a future plan relayed by an aggregator, not a launch confirmed here.

Watch for details from Airbnb on what the agent will do and which AI models it will use.

Feature tracker TestingCatalog says a Slack integration for Grok Bot is rolling out or about to. Users would connect their bots to Slack individually and trigger them through direct messages or mentions, tied to a forthcoming macOS build. Release timing and availability are not confirmed.

Watch for an official announcement and which plans get the integration.

TestingCatalog says Anthropic is working on a collaborative mode for Claude Code Projects, and that a recent activity view on the Projects page summarizes the latest threads. It quotes Lydia Hallie of Anthropic saying the day before that "we don't have multiplayer yet." This is an anticipated feature, not a shipped one.

Watch for an official Anthropic announcement of shared projects.

TestingCatalog recaps an upgrade to Mistral Studio from earlier this week. It says users can now build and host AI apps and improve them from real usage, with connectors, knowledge, a search index and a secrets manager set up once and reused across projects, plus agents, workflows, app deployment and durable execution. The official Mistral announcement was not fetched.

The AI 2027 tracker compares real progress with the AI 2027 forecast. kimmonismus says its latest update, based on scores from METR's report on Anthropic's internal models, puts some of those models ahead of the forecast's curve. Andrew Curran says progress went from five weeks behind the curve to six weeks ahead. These are commentators' readings of the tracker, not confirmed findings.

Polymarket says TikTok parent ByteDance is reportedly considering a massive expansion of its AI data centers in Brazil. The post names no original reporting source, so treat the claim as unverified.

Watch for reporting from an outlet that names its sources, or a statement from ByteDance.