Top stories
- A Meta AI security researcher’s viral X post described an OpenClaw agent that “ran amok” in her inbox — a cautionary example of what can go wrong when autonomous agents act on behalf of users. (TechCrunch)
- Anthropic publicly accused DeepSeek and other Chinese firms (including Moonshot and MiniMax) of large-scale campaigns — allegedly using ~24,000 fake accounts — to distill and mine Claude’s capabilities. (TechCrunch / The Verge)
- Google Cloud argues models are advancing on three frontiers at once: raw intelligence, response time, and extensibility — and positions Cloud AI to lead across them. (TechCrunch)
- At least a dozen investors that backed OpenAI also now back Anthropic, highlighting shifting loyalties and conflict-of-interest concerns in AI VC circles. (TechCrunch)
- Tesla filed a lawsuit against the California DMV, extending its ongoing legal fight over Autopilot rules and enforcement. (TechCrunch)
- Uber Autonomous Solutions is positioning itself as a full-stack vendor for robotaxi and delivery fleets — selling software, services and operations support. (TechCrunch)
- Ex-Apple team launched Acme Weather, a consumer-first weather app with alternative forecasts, rainbow and sunset alerts, and other product twists. (TechCrunch)
Key takeaways
- Treat autonomous agents like production services: enforce least privilege, verification steps, and audit logs before delegating inboxes or accounts.
- Model IP and training hygiene are rising flashpoints — expect more public accusations and regulatory scrutiny around data-mining and distillation.
- Platform-strength is now multi-dimensional: latency, capability, and extensibility matter when choosing cloud AI providers.
- Investor alignments in AI are fluid; dual investments may reshape governance and perceived conflicts of interest.
- Industry players keep expanding: automakers, fleets, and startups are all staking claims in the next-gen mobility and consumer app spaces.
Why it matters
Autonomous-agent convenience collides with security and governance gaps, while model-extraction accusations and shifting investor loyalties signal sharper legal, technical and policy battlegrounds ahead. Product moves from Google, Tesla, Uber and new startups show firms are betting on differentiated platform capabilities and vertical integration.
Sources
- A Meta AI security researcher said an OpenClaw agent ran amok on her inbox (TechCrunch) — https://techcrunch.com/2026/02/23/a-meta-ai-security-researcher-said-an-openclaw-agent-ran-amok-on-her-inbox/
- Anthropic accuses DeepSeek and other Chinese firms of using Claude to train their AI (The Verge) — https://www.theverge.com/ai-artificial-intelligence/883243/anthropic-claude-deepseek-china-ai-distillation
- Anthropic accuses Chinese AI labs of mining Claude as US debates AI chip exports (TechCrunch) — https://techcrunch.com/2026/02/23/anthropic-accuses-chinese-ai-labs-of-mining-claude-as-us-debates-ai-chip-exports/
- Google’s Cloud AI leads on the three frontiers of model capability (TechCrunch) — https://techcrunch.com/2026/02/23/googles-cloud-ai-lead-on-the-three-frontiers-of-model-capability/
- With AI, investor loyalty is (almost) dead: At least a dozen OpenAI VCs now also back Anthropic (TechCrunch) — https://techcrunch.com/2026/02/23/with-ai-investor-loyalty-is-almost-dead-at-least-a-dozen-openai-vcs-now-also-back-anthropic/
- Tesla’s battle with the California Department of Motor Vehicles isn’t over after all (TechCrunch) — https://techcrunch.com/2026/02/23/teslas-battle-with-the-california-department-of-motor-vehicles-isnt-over-after-all/
- Uber wants to be a Swiss Army Knife for robotaxis (TechCrunch) — https://techcrunch.com/2026/02/23/uber-autonomous-solutions-av-robotaxi-delivery-robots/
- Ex-Apple team launches Acme Weather, a new take on weather forecasting (TechCrunch) — https://techcrunch.com/2026/02/23/ex-apple-team-launches-acme-weather-a-new-take-on-weather-forecasting/
Not financial/professional advice