AI News, Aug 23, 2026: Inherent releases AI agent to replicate scientific research
The Brief
Sunday, August 23, 2026
Today's updates show AI moving deeper into specialized roles, from replicating scientific research to powering cybersecurity and education.
- 01
Harvard's $699 HBS Foundry startup program is using AI avatars of its instructors to provide students with feedback during practice sessions.
Why it matters
Students can now practice pitches and board meetings with interactive AI personas, making personalized business coaching more accessible.
- 02
British AI startup Inherent released Faraday, an AI agent designed to replicate scientific papers, which reportedly outperformed models from OpenAI and Anthropic.
Why it matters
This tool could speed up scientific discovery by automating the tedious process of verifying and reproducing published academic papers.
- 03
OpenAI has called on California lawmakers to strengthen SB 53, an AI safety bill that the company had previously opposed.
Why it matters
The shift in stance from a major AI developer could influence how future AI safety and compliance laws are written and enforced.
- 04
Netflix is testing an in-house language model called GenRec that converts viewing history into plain text to generate better content recommendations.
Why it matters
Replacing complex, hand-built algorithms with text-based AI could make your streaming recommendations more accurate and personalized.
- 05
Anthropic released Claude Security, a tool powered by its Claude Mythos 5 model that scans codebases for vulnerabilities and suggests patches.
Why it matters
Organizations can use this tool to automatically scan codebases, find vulnerabilities, and get suggested patches to protect critical infrastructure.