Radar Trends is bursting with relevant news for July, including some of the latest security gaps. Here’s a bit of what Mike Loukides’s sleuthing has turned up.
Is AI going the way of Web 2.0 in its readiness to erect barriers? We’re looking at you, Anthropic, Slack, and Google. A new breed of “expert generalists” is leaning away from tech specialization and into a more inclusive skill set. To limit hallucinations, an MCP server lets confused AI agents request clarity from the human in the loop. And a new nonprofit AI research group will attempt to create AI without antisocial tendencies toward “deception, self-preservation, and goal misalignment.” Finally, how many qubits does it take to break a 2,048-bit RSA encryption key? Fewer than you may think.