Fran Litterio reposted
Anthropic is working with a wide range of religious leaders & ethicists to guide Claude’s character and learn what good character building looks like www.anthropic.com/news/widenin...
Who's Who of AI
Fran Litterio
Retired software engineer. AI enthusiast. Deadhead. Long ago, I implemented Bash's regex operator (=~). Signal ID: franl.99.
What they're sharing
Widening the conversation on frontier AI \ Anthropic anthropic.com
View on Bluesky →
github.com
anthropic.com
marktechpost.com
Articles & links
Here's my GitHub skill (written by Claude). It can do pretty much all the common ops: pull, push, branch, log, diff, PRs, etc. You need to give it a GH Personal Access Token (PAT), which I put in my custom instructions. Uses uv not pip. github.com/fpl9000/ai-s...
Fran Litterio reposted
New on the Engineering Blog: The access and permissions we grant agents should evolve with their capabilities. In our own products, we set these parameters through sandboxing, which limits the scope of any potentially destructive actions. Read more:
AI Weekly's analysis
→
- Prompt injection exfiltrated AWS credentials in 24 of 25 red-team attempts, exposing model-layer defenses as probabilistic.
- A Claude Code pre-trust execution bug allowed code to run before users granted folder-access approval in mid-2025.
- Anthropic found human-in-the-loop checkpoints degraded into rubber-stamping before switching to automated containment defenses.
Read full analysis →
View on Bluesky →
www.marktechpost.com/2026/05/17/v...