AI News Brief: Morning roundup, 2026-11-22
Summary
The period was marked by AI systems running into real-world consequences. South Korea's education minister resigned after AI-assisted cheating disrupted the run-up to university entrance exams, Tesla began rolling out a major Full Self-Driving update, and the Red Cross warned that rules for autonomous weapons are falling behind. The common thread is accountability: who oversees these systems in classrooms, cars and battlefields. Model behavior is under scrutiny too, with a comparison finding OpenAI's GPT-6.2 variants less willing to admit uncertainty than Anthropic's Claude. Together, the stories suggest that governance and trust are struggling to keep pace with deployment.
- Google launched Tool Retrieval in preview for the Gemini API, letting agents keep a large tool library on standby and load each tool's parameter schema only when it is needed, all within one Interactions API request. It is aimed at developers building agents that must choose among many tools.
- OpenAI's GPT-6.2 Astra and Sol give much shorter, more direct answers than their predecessors, but they are less willing to admit uncertainty. In the comparison reported by The Decoder, Anthropic's Claude models score well ahead on transparency.
- The International Committee of the Red Cross and rights groups warned that it may soon be too late to put the "genie back in the bottle" on autonomous weapons, as a UN conference failed to agree on a regulatory instrument. Development appears to be outpacing efforts to regulate it.
- Tesla began rolling out Full Self-Driving v15 on November 21, adding an upgraded driving network, a live Reality View display and changes to navigation and parking. The company reports a 25-fold improvement in its critical-intervention rate.
- Berlin's Kraftwerk, part of the same development as the techno club Tresor, was turned into a "war room" for an arms firm's event. The move drew outrage over military AI being showcased at a cultural venue.
- South Korea's education minister, Choi Kyo-jin, resigned after an AI-assisted cheating scandal engulfed the university entrance exams. He said he had "failed to prepare adequately," even though this year's exams banned phones and checked for cameras hidden in clothing.
Summaries written with AI from the linked source articles.
Sources in this roundup
| AI (artificial intelligence) | The Guardian | 3 article(s) | |
| MarkTechPost | 1 article(s) | |
| The Decoder | 1 article(s) | |
| Unite.AI | 1 article(s) |
Most-mentioned keywords
| tool | 2 mention(s) | |
| api | 1 mention(s) | |
| arts | 1 mention(s) | |
| astra | 1 mention(s) | |
| autonomous | 1 mention(s) | |
| berlin | 1 mention(s) | |
| cheating | 1 mention(s) | |
| claude | 1 mention(s) |
Sources
- Google AI Previews Tool Retrieval in the Gemini API for Large Tool Libraries
- GPT-6.2 Astra and Sol get to the point faster but score lower on transparency than Claude
- It will soon be too late to regulate use of autonomous weapons, warns Red Cross
- Tesla Rolls Out FSD V15 with Reality View and Vision Upgrades
- ‘Serious error of judgment’: outrage over military AI event at Berlin arts complex
- South Korea education minister resigns after AI cheating scandal engulfs university exams
