So an AI can pick up its identity, which it then models its behaviour on, through the text in its dataset. The huge corpus of classic evil AI sci-fi novels and misaligned AI research papers in the datasets of GPTs could be a serious issue for AI labs: www.transformernews.ai/p/why-ai-rea...

There are already hints this could be an issue. In some evaluations, Sonnet 4.5 explicitly mentioned being tested over 80% of the time, compared to 10% for Sonnet 4. What caused this jump? It was likely due to exposure to misalignment evaluations during training.

The preemption fight is certainly not over — but the position of the groups wanting to regulate AI has never been stronger, thanks in part to their opponents’ repeated miscalculations.
You can read more about the ongoing battle for preemption here: www.transformernews.ai/p/another-pr...

President Trump endorsed the preemption efforts shortly after a White House visit from the head of an industry-backed group. Since that endorsement has possibly made Trump look weak and not in control of his party, he might be wary of taking advice from such people in future.

The AI industry pushing against regulation finds itself on the backfoot in other ways too — AI is becoming increasingly unpopular in the US, and as time passes, politicians will become more worried about giving AI companies a free pass.

www.transformernews.ai/p/will-ai-sa... So could AI safety go from an insular concern to a mass political movement? And should it?

“Back in 1970 the correct move wasn’t coming up with a foolproof plan to solve climate change” says Erik Hoel “the correct move was activism. AI safety advocates should therefore look to climate activists to see what’s effective. Which is basically panic, lobbying, and outrage.”

There's little cohesion among broader anti-AI sentiments, between Hollywood writers going on strike over AI clauses, local communities protesting data centers, and researchers producing reports on the singularity. But a mass movement on a focused targets could tie it all together

The reason this hodgepodge works, she explains, is that there is agreement on an overall goal: to slow and reverse the level of heating. There are many disagreements within the movement, but this unified purpose makes it coherent enough to have mass appeal.

“One of the key learnings, if you like, is about strategic and tactical diversity,” she explains. Some parts of the movement focus on changing public thinking on a grand scale; others push for specific changes, using everything from in-person protest to electing green candidates

SB 53 forces California's AG to address this problem, and the AG’s office has already announced that it is hiring an AI expert. When regulators take whistleblower protections seriously, companies change internal policies for the better. Read the full op-ed: www.transformernews.ai/p/sb-53-prot...

SB 53 will necessitate California’s civil service gets to grips with AI. If a whistleblower reports that a frontier model could assist in creating bioweapons or has escaped control, the office has to understand what that means, test whether it’s true, and figure out what to do.

The bill’s most important impact might however be an unintended one — it could force California’s government to finally build AI expertise. Currently, that expertise barely exists, preventing insiders from raising concerns.

The new SB 53 places a legal responsibility on frontier AI labs to report whether their models pose a risk to the population. But it places a far greater moral responsibility on their employees to say what those risks are, rather than outside investigators.