A federal judge handed Meta a win in a major copyright case over using books to train AI models. But the decision wasn’t exactly a validation of Meta’s practices. It was a result of the authors failing to argue their case effectively.
Judge Vince Chhabria ruled in Meta’s favor after finding that the authors who sued didn’t present the right arguments or evidence. They claimed Meta’s Llama models let users reproduce their book content and that Meta harmed the market for licensing books to AI companies. Chhabria dismissed both arguments. He said the AI model couldn’t reproduce long excerpts even with aggressive prompting, and that authors don’t have the right to control the entire market for AI training licenses. [Read more…]
The buzz around AI-powered coding tools is hard to avoid, and I get asked about them a lot. Instead of continuing to repeat myself I figured it was time to write down exactly why I don’t use them, and why that decision isn’t about being for or against AI in some grand ideological sense. This is just a practical take based on how I work, what I value, and what actually helps me get things done.
Email authentication continues to improve, but DMARC is still widely misunderstood and often misused. The number of domains publishing DMARC records has grown steadily since 2023, especially in industries like finance, education, and healthcare. However, a large percentage of those domains are not enforcing any policy, which means spoofing and phishing attacks are still a serious concern.
Security researchers have identified two major exploits in the Secure Boot system, both capable of sidestepping one of the most important protections on modern PCs. Microsoft has issued a patch for one of them. The other remains untouched, even as it offers attackers a nearly universal method to bypass security during the startup process.
OpenAI Faces Pressure to Store All ChatGPT Conversations