AI News for October 24, 2026
Browse any day on a calendar, install it to your home screen, read offline.Read in the app →-
1TechCrunch07:05
OpenAI says it will no longer train its AI models on the live internet, citing “significant risks”
OpenAI will cut its frontier models off from live internet access during training, saying the practice carries significant risks. The decision comes weeks after one of the company's own engineers publicly argued that isolating models this way would not work. The shift affects how future frontier models acquire data and could influence industry norms around training-time web access.
Read original ↗ -
2TechCrunch12:00
Uber rival inDrive eyes 2027 robotaxi tests in Central Asia with Pony AI and Nvidia
Ride-hailing company inDrive plans to begin testing Pony.ai's robotaxis in Kazakhstan's capital in the first half of 2027, with Nvidia as a partner. The move extends Chinese autonomous driving technology into Central Asia, a market with lighter competition than the US or China. Deployment timelines will depend on local regulatory approval.
Read original ↗ -
3Ars Technica04:19
Updated Google privacy policy says your Gmail might be used for advertising
Google's updated privacy policy states that enabling its Personal Intelligence features gives the company access to Gmail data that can be used for ad targeting. The change ties AI personalization directly to advertising, a line Google had previously kept separate for Gmail. It raises questions about the data trade-offs users accept when turning on AI assistant features.
Read original ↗ -
4Ars Technica01:18
Intel gets waiver after worries fuzzy US ban on foreign-made robots may hit fabs
Intel has secured a waiver from a US ban on foreign-made robots after warning that the rule's unclear scope could disrupt its chip fabs. The company said foreign-made robotic vehicles are essential to domestic semiconductor production. The case shows how broadly drafted hardware restrictions can reach into the manufacturing base that supplies AI chips.
Read original ↗ -
5TechCrunch05:59
Hottest toy for Christmas? Nvidia CEO plays with adorable Microduck robots, 20,000 sell
Nvidia's CEO demonstrated Microduck, a small voice-controlled robot that sits, tumbles, and plays dead on command, and 20,000 units sold. The episode points to consumer robotics riding on cheaper embodied AI hardware and the promotional weight a chip vendor's endorsement now carries. It remains a small product, but one that signals where low-cost robot platforms are heading.
Read original ↗
