Oh Cool, AI Is Inventing New Biological Viruses Now
THE SO WHAT
If AI systems can meaningfully assist in designing new biological agents, the risk surface for labs, cloud providers, and model operators shifts from “misuse of outputs” to “dual-use R&D pipeline.” Expect tighter access controls, logging, and domain-specific guardrails around bio capabilities — and budget time for compliance reviews if your models or infra touch life sciences.
READ THE SOURCE
MORE FROM THE WIRE
Applied AINo cloud, no GPUs, no problem: Liquid AI's new model LFM2.5-2.6B brings powerful AI agents to devices as small as a Raspberry Pi
Agentic workloads dropping onto Raspberry Pi-class hardware means some automation will bypass cloud and MDM entirely. If you run field devices or edge fleets, you now have to plan for on-device agents as both a capability and a security surface.
Applied AIOpenAI’s new AI smart speaker will reportedly sell for between $300-$400
A $300–$400 OpenAI smart speaker is a shot at owning the high-intent, always-on assistant surface in the home. If your product depends on mobile or browser entry points, assume some share of queries and transactions will be mediated by a model-first device instead.
Applied AISources: Canva slashed revenue growth forecast as heavy use of new AI features drove up costs and slowed their rollout, while more Canva users turned to ChatGPT
Canva’s AI features driving up infra costs while users defect to ChatGPT is the downside of bolting expensive models onto a freemium base without a clear monetization or retention wedge. If you’re adding AI to a mass-market product, model cost, usage caps, and substitution risk need to be designed into the business model, not patched after launch.
Applied AIQuote of the day by computer scientist Geoffrey Hinton on AI: 'It is hard to see how you can prevent the bad actors from using it for bad things' — a pessimistic take on the domination of new technologies by those with ulterior motives
Hinton’s line about bad actors is a reminder that AI risk is now assumed, not hypothetical, in board and regulator conversations. Teams deploying advanced models should expect more scrutiny on abuse cases and be ready with concrete mitigations, not just productivity narratives.