1. Mythos found 271 zero-days in Firefox 150. Mozilla says developers are in for a rough few years anyway. The Firefox 150 codebase went through something new this quarter: a full-coverage scan by Anthropic's Mythos model.

2. Atlassian flipped customer tickets to AI training by default. Meta will log every employee keystroke. Same week. Two consent defaults flipped in seven days. Atlassian enabled customer data collection for AI model training by default, leaving admins to find and turn off the opt-out toggle themselves.

3. A med student built a MAGA influencer who doesn't exist, and says "super dumb" men paid him thousands A medical student told Wired he has earned thousands of dollars selling photos and videos of a young conservative woman he fabricated entirely with generative tools. The character does not exist.

In Brief

• Anthropic takes $5B from Amazon, commits $100B back to AWS Amazon invested $5 billion in Anthropic, and Anthropic agreed to spend $100 billion on AWS compute over the term. The circular flow ties Claude training capacity directly to Amazon's cloud revenue line.

• SpaceX structures $60B Cursor deal with $10B walk-away fee SpaceX announced an option to acquire Cursor for $60 billion or pay a $10 billion break fee, ahead of the planned SpaceX/xAI/X combined IPO. The structure folds an AI coding tool into Musk's pre-IPO portfolio without committing to close.

• Tesla hid thousands of fatal autonomous-driving incidents, Swiss outlet reports RTS reports Tesla concealed thousands of incidents involving its autonomous driving systems, including fatal ones, to keep testing on public roads. The disclosures arrive while regulators in multiple jurisdictions weigh approval expansions.

• OpenAI hits 4M weekly Codex users, signs Accenture, PwC, Infosys OpenAI launched Codex Labs and named Accenture, PwC, and Infosys as deployment partners for enterprise rollouts. Weekly Codex users reached 4 million, giving OpenAI scale data on how integrators package coding agents for Fortune 500 buyers.

• OpenAI ships ChatGPT Images 2.0 OpenAI released a new version of its in-chat image generation, with updated rendering and editing inside ChatGPT. The launch follows competitive releases from Google and Black Forest Labs over the past month.

• Clarifai deletes 3M OkCupid photos under FTC settlement Clarifai deleted 3 million OkCupid user photos it had used to train facial recognition models, complying with an FTC settlement. Court documents show OkCupid executives invested in Clarifai before sharing user data in 2014.

• YouTube opens deepfake takedown tool to celebrities YouTube extended its likeness detection program to Hollywood public figures, letting enrolled celebrities scan the platform for AI-generated videos of themselves and request removal. The tool was previously limited to creators in YouTube's Partner Program.

• NeoCognition raises $40M seed for human-style learning agents Founded by an Ohio State researcher, NeoCognition closed a $40 million seed to build agents that acquire domain expertise through continual learning rather than fine-tuning. The check size is large for a pre-product seed in agent research.

• The Verge: Starbucks ChatGPT ordering app fails on a standing order A Verge reporter tried Starbucks' ChatGPT-powered ordering app for a drink she has ordered for years, and the bot mishandled the request. The review documents friction in voice-driven QSR ordering, a category multiple chains piloted this year.

• The Verge: AI is missing from US midterm campaign agendas despite voter concern US polling shows broad public concern about AI, opposition to data center projects, and online anger at AI executives. Campaign messaging for the 2026 midterms has not picked up the issue, leaving a gap between voter sentiment and stated policy priorities.

• Bond launches AI social app aimed at reducing scroll time Bond opened a social platform whose AI prompts users toward offline activities and logs them as memories rather than feed posts. The product positions itself against engagement-maximizing feeds at a moment when several US states are weighing minor-use restrictions.

Read the full edition →