Daily AI industry insights for professionals
OpenAI is calling for California to strengthen SB 53, an AI safety bill the company previously opposed, according to TechCrunch AI.
Why it matters: OpenAI now supports strengthening California's SB 53 after previously opposing the bill, reversing the company's stated position on the legislation.
Anthropic is running its Claude Security scanner on Claude Mythos 5, its most powerful model. The tool scans codebases for vulnerabilities, assigns severity ratings with CWE classifications, and suggests patches, according to The Decoder.
The deployment puts Mythos 5 inside an operational cybersecurity product rather than leaving it as a general-purpose model. The report provides no benchmark results, pricing, deployment scale, or comparison with competing scanners.
Why it matters: Anthropic now has Claude Security using Mythos 5 to produce CWE-classified findings and patch suggestions, but the supplied report does not show whether the scanner detects or fixes vulnerabilities better than competing tools.
AI agents have consumed more tokens than humans on OpenRouter since February 6, 2025. The Decoder AI reports that agentic usage has grown 14x since then, compared with 2.8x growth in human usage.
Nearly 70 percent of agent token consumption comes from cheap cached prompts. The Decoder AI nevertheless reports that actual costs are rising as agent activity expands.
Why it matters: OpenRouter now processes more agent than human tokens, and rising actual costs show that the shift is increasing spend even with cached prompts making up nearly 70 percent of agent consumption.
Cheap energy, abundant land, and proximity to Beijing have turned an unnamed city in Inner Mongolia into a crucial hub for China’s data centers, according to Wired AI.
Wired AI leaves the city’s name, data-center capacity, investment total, and energy costs unspecified. The report identifies the location’s three advantages but gives no measure of the buildout’s scale.
Why it matters: For data-center operators choosing sites in China, Wired AI links cheap energy, abundant land, and proximity to Beijing to a crucial Inner Mongolian hub—while providing no capacity or investment figures to size the opportunity.
Waymo has doubled its lobbying spending in a robotaxi battle with Uber. The Alphabet-owned company is seeking to persuade U.S. regulators to clear a path for fully autonomous taxi services.
Why it matters: The robotaxi fight with Uber now has a lobbying front: Waymo is spending more to pursue regulatory clearance.
Tesla says Cybercab will launch in 12 days and begin entering robotaxi fleets in Austin and elsewhere. The company posted the update on X on August 22.
Tesla's post also referenced future announcements about rollout schedules and plans, but did not provide those details.
Why it matters: Tesla's 12-day claim puts Austin robotaxi fleets on a near-term clock, while the company's post points to additional rollout schedules and plans still to come.
Google DeepMind is partnering with game studios to prototype AI gameplay systems.
The effort builds on 15 years of game-AI research, from Atari to EVE Online, according to the project’s source article.
Why it matters: Game studios now have Google DeepMind as a research partner for prototyping AI gameplay, connecting the lab’s 15 years of game research to studio development.