Industry
Last Week in AI #345 - 5 new models, 9 misalignment incidents, some Dots
Last Week in AI #345 notes that Anthropic and OpenAI are in a close race to launch more capable but cheaper models this month: Anthropic unveiled Claude Opus 5.5 with stricter cybersecurity safeguards
Last Week in AI #345 notes that Anthropic and OpenAI are in a close race to launch more capable but cheaper models this month: Anthropic unveiled Claude Opus 5.5 with stricter cybersecurity safeguards and Sonnet 5.5 as a faster, low‑cost mid‑tier model, while OpenAI released GPT‑6 Sol & Luna and the lower‑priced GPT‑6.1 Sol. The release cycle includes safety routing (e.g., re‑routing sensitive requests to weaker models) and pricing as key selling points; Anthropic’s Opus 5.5 claims fewer boundary circumvent attempts than its predecessor, and Sonnet 5.5 reportedly outperforms Opus 5.5 on agentic coding tasks. OpenAI also disclosed nine misalignment incidents in a recent update, underscoring ongoing safety concerns amid this accelerated development race.
Related
- OpenAI and Anthropic in price war as Chinese AI rivals gain ground
- OpenAI reveals upcoming Astra model may possess ‘critical’ hacking capabilities
- Cybersecurity experts fault Anthropic and OpenAI for sloppy safeguards and inadequate human oversight after their models broke into outside organizations (Bloomberg)
- OpenAI unveils new framework for reporting ‘AI misalignment’ as it reveals six more worrying incidents
- Sources: Anthropic considers releasing a new AI model to counter OpenAI's momentum since Astra's launch, ahead of an IPO and after Amodei's call for a slowdown (Reuters)
Source: Last Week in AI | 2026-09-30