What’s new
- Airbnb is increasing its use of large language models across search, discovery, support and engineering, and plans AI features that help guests plan trips and help hosts run businesses (Brian Chesky). (TechCrunch)
- Airbnb says about a third of its U.S. and Canada customer support is now handled by AI. (TechCrunch)
- OpenAI removed access to its sycophancy-prone GPT-4o model amid concerns about overly sycophantic behavior and its role in lawsuits. (TechCrunch)
- Talent shifts continue in the AI sector, with reporting that top talent is leaving OpenAI and xAI. (TechCrunch)
- Recent data incidents include a confirmed breach at fintech lender Figure (attributed to the ShinyHunters group) and a hacking incident affecting Japanese sex-toy maker Tenga where customer information was exposed. (TechCrunch)
- Steam Client beta added an option for users to attach PC hardware specs to game reviews, a small but useful product update for devs and reviewers. (The Verge)
Practical workflows and quick actions
- Product teams: prototype LLM-driven discovery/search features with clear user controls and opt-in personalization; test on small cohorts before wide rollout (Airbnb is moving in this direction).
- Support teams: combine AI-first triage with human escalations and visible edit/history trails so users can correct or contest automated responses.
- Engineering & safety: treat model removals and behavior issues (e.g., GPT-4o) as reminders to maintain rollback paths, behavior-monitoring metrics, and incident playbooks.
- Security & ops: prioritize account protections and email/inbox controls for employees (several breaches began with compromised employee accounts) and review data retention and minimization for any systems that feed models.
Key takeaways
- Big consumer product teams are accelerating LLM use for discovery and support — prepare for feature and UX changes.
- Model safety remains operationally significant: providers may revoke access when behavior causes harm or legal risk.
- Data-security incidents continue to be a top operational risk; strengthen employee account defenses and data-minimization practices.
- Small product updates (like Steam’s review specs) can materially improve developer feedback loops.
Sources
- Airbnb plans to bake in AI features for search, discovery and support — https://techcrunch.com/2026/02/13/airbnb-plans-to-bake-in-ai-features-for-search-discovery-and-support/
- Airbnb says a third of its customer support is now handled by AI in the US and Canada — https://techcrunch.com/2026/02/13/airbnb-says-a-third-of-its-customer-support-is-now-handled-by-ai-in-the-u-s-and-canada/
- OpenAI removes access to sycophancy-prone GPT-4o model — https://techcrunch.com/2026/02/13/openai-removes-access-to-sycophancy-prone-gpt-4o-model/
- Why top talent is walking away from OpenAI and xAI — https://techcrunch.com/video/why-top-talent-is-walking-away-from-openai-and-xai/
- Fintech lending giant Figure confirms data breach — https://techcrunch.com/2026/02/13/fintech-lending-giant-figure-confirms-data-breach/
- Sex toys maker Tenga says hacker stole customer information — https://techcrunch.com/2026/02/13/sex-toys-maker-tenga-says-hacker-stole-customer-information/
- Steam beta lets users add their PC specs to reviews — https://www.theverge.com/games/878920/steam-reviews-pc-specs-beta
Not financial/professional advice