Roxhall
banner
roxhall.com
Roxhall
@roxhall.com
We build systems that build businesses.
Pinned
Ok, so toddlers possess Gifted Word Learner skills. But they still can't herd sheep!
January 22, 2026 at 3:04 AM
Rug pulls like Anthropic's are par for the course. Experienced engineers know this from years of watching platforms evolve. medium.com/ai-software-...
Anthropic Just Pulled the Rug on Competition (Locked Models to Claude Code Only)
Anthropic has just changed the rules of the game and caught everyone off guard — including those building tools on top of their models.
medium.com
January 22, 2026 at 12:29 AM
Whether Witness AI or competitors like Robust Intelligence and Protect AI win out, the underlying need for AI security will only increase. Organizations require greater visibility and control layers as AI tool proliferation continues. It's not just a compliance play, it's a core operational need.
January 19, 2026 at 5:58 PM
The Atlantic's portrayal oversimplifies the issue and, in our view, fuels more controversy than the topic deserves. Memorization is a manageable edge case, not a core flaw. Companies are tackling this w/data filtering, alignment techniques, safeguards & licensing. www.theatlantic.com/technology/a...
AI’s Memorization Crisis
Large language models don’t “learn”—they copy. And that could change everything for the tech industry.
www.theatlantic.com
January 17, 2026 at 5:27 PM
Ok, so toddlers possess Gifted Word Learner skills. But they still can't herd sheep!
January 10, 2026 at 3:32 PM