Yo, check this out. Trump just signed an executive order scaling back the AI cybersecurity stuff.<br> <br> So, the gist is, the government is finally putting a framework in place to actually *test* new AI models before they hit the public. The deal is that companies sharing their most powerful models have to let the federal government review them for vulnerabilities—especially for critical infrastructure like banks and hospitals—before the models get released. The original plan had a whopping 90 days, but they scaled it way back to a maximum of 30 days.<br> <br> This is actually a pretty big deal. For a while, it felt like the administration was more interested in ideological control over AI (like trying to ban "woke" models) than actual security risks. This move signals that Trump and his team finally realized AI poses real security threats, and they need to act.<br> <br> The main takeaway here is the move from laissez-faire to actual regulation. It’s a step in the right direction, even if the process is still a little "opaque," as some folks pointed out. It’s good that they’re forcing a teeth into the process instead of just letting the big players roll out the next shiny thing without a safety net.<br> <br> What do you guys think? Is 30 days enough time, or is this just a cosmetic fix?<br> <br> Source: https://www.engadget.com/2185845/trump-signs-scaled-back-ai-cybersecurity-order/