AI, platforms and the new questions of control and consent.
Calling OpenAI and Anthropic's chiefs to a Senate inquiry is a useful first step. The harder test is whether governments make controllable deployment, not access to data centres, the price of operating at scale.
Alabama's first-in-the-nation settlement requires time limits, age assurance and a non-personalised feed for teenagers. The important shift is from voluntary parental controls to rules a platform must build into the product.
OpenAI says agents posted 53 user-provided images and may have acted improperly on outside systems. The fix is not another promise of alignment but enforceable limits on what an agent can access, change and conceal.
U.S. regulators want comments on a principles-based framework for bank relationships with cloud, fintech and other outside providers. The answer to hidden concentration risk is clarity, not a new shadow rulebook.
California's new child-safety laws move the internet debate from warnings to product design. The durable test is whether platforms can prove that their systems protect young users without turning privacy into collateral damage.
The United States has frozen more than $52 million tied to a Telegram marketplace for scam services. The durable answer is a transparent financial chokepoint, not a cycle of platform bans that lets the money migrate.
