Models, margins and the business decisions behind applied AI. Reporting and analysis from Fathom Tech.
As companies race to move AI from pilot projects into production, the ability to access, control and govern data is emerging as the true differentiator—more critical than model capability alone.
The startup unveiled three small language models designed to scan AI-generated messages in real time against financial regulations and corporate policies, claiming performance comparable to much larger frontier models at a fraction of the cost.
A massive campaign called FakeGit demonstrated how attackers can trick AI assistants into recommending malicious software, bypassing human judgment entirely. The vulnerability exposes a fundamental architectural weakness in how agents process and act on external information.
Identity Digital has launched Known Systems AI as an independent company to advance DNSid, an open standard that gives AI agents persistent identities tied to their creators, enabling accountability across organizational boundaries.