Diffusion Large Language Models (dLLMs) break the rigid left-to-right constraint of traditional LLMs, enabling token generation in arbitrary orders. Intuitively, this flexibility implies a solution space that strictly supersets the fixed autoregressive trajectory, theoretically unlocking superior reasoning potential. Indeed, for specific constraint satisfaction tasks (e.g., sudoku puzzles), this capability has proven to be highly advantageous. However, in this paper, we reveal that for general reasoning tasks (e.g., mathematics and coding), arbitrary order generation may in fact limit the reasoning potential of dLLMs. We find that dLLMs tend to exploit this order flexibility to bypass high-uncertainty tokens that are crucial for exploration, leading to a premature collapse of solution coverage. This observation motivates a rethink of RL approaches for dLLMs, where considerable complexities, such as handling combinatorial trajectories and intractable likelihoods, are often devoted to preserving this flexibility. We demonstrate that effective reasoning can be better elicited by simply forgoing arbitrary order and applying standard Group Relative Policy Optimization (GRPO) instead. Our approach, JustGRPO, is minimalist yet surprisingly effective (e.g., 89.1% accuracy on GSM8K) while fully retaining the parallel decoding ability of dLLMs.

'Nobody is falling for a toucan sighting in Siberia' This image is AI-generated, using ChatGPT. 'Nobody is falling for a toucan sighting in Siberia' This image is AI-generated, using ChatGPT. For many birdwatchers, recording a species outside its normal range is the holy grail. In the UK, the discoveries often make national headlinesThe western reef heron, for example, usually found in Africa and southern Europe, spotted in a seaside town in north Wales in June, which was widely celebrated on birding forums. But a new scourge is threatening to disrupt the fun: AI slop.

"Context bombing" tricks malicious AI agents into shutting down before they can do harm. Prompt injections, the malicious commands attackers embed into content to entice large language models to follow them, have been attackers' go-to tool for turning AI platforms against their users. A well-phrased command sneaked into an email or calendar invitation is often all it takes to cause the LLM to exfiltrate sensitive data or follow other harmful actions. Now, defenders are embracing the prompt injection, too. Researchers from Tracebit on Monday said they found that placing prompt injections alongside passwords, cryptographic keys, and other secrets stored on Amazon Web Services was often all that was needed to shut down attacks from AI hacking agents.

Follow this section to personalize your feed and get instant alerts. Follow Go to your personalized feed WHY FOLLOW? Smart Alerts: Get notified about major news as it happens. Follow this tag to personalize your feed and get instant alerts. Follow Go to your personalized feed WHY FOLLOW?

Engadget Podcast: Is Siri AI actually useful in iOS 27 and macOS Golden Gate? Engadget Podcast: Is Siri AI actually useful in iOS 27 and macOS Golden Gate? Also, we try to make sense of OpenAI's rumored AI smart speaker. Apple finally launched the public betas for iOS 27, macOS Golden Gate and all of its other new software this week. Siri AI seems genuinely useful, and potentially less annoying to consumers than Microsoft's Copilot.

The City Attorney's Office sent the tech giants cease-and-desist letters this week telling them to stop profiting from 13 "face-swap" apps that are overwhelmingly used to target women and girls. Apple and Google have been ordered to take down apps that can "nudify" or "undress" people and told that they must stop profiting from the harmful technology, according to cease-and-desist letters sent to the companies seen by WIRED. On Thursday, San Francisco city attorney David Chiu sent legal notices to Apple and Google demanding that they remove from their app stores 13 face-swapping apps, which allow users to create AI-generated nonconsensual nude images. The letters say the Silicon Valley giants should stop "aiding and abetting" the sale of explicit deepfake images and "sever" business relationships with the app developers. "Generating non-consensual intimate images is illegal, harmful, and completely unacceptable," Chiu tells WIRED.

Google Search AI poses'unacceptable risk' to kids, report finds A new report from a California-based nonprofit says the artificial intelligence features built in to Alphabet's Google Search pose an unacceptable risk" to children. The artificial intelligence features built in to Alphabet's Google Search pose an "unacceptable risk" to children, according to a new report from the Youth AI Safety Institute at Common Sense Media, a California-based nonprofit. The study found that on test accounts configured for minors, the ubiquitous search engine's AI features failed to detect suicide risks, said an eating disorder symptom was normal, and provided instructions for creating deepfakes, or sexually explicit fake content. The report added that unlike standalone chatbots, Google's AI tools are built directly into the default search experience on school and personal devices, currently offering administrators or parents no way to disable them. The report specifically evaluated AI Overviews, which generate automated answers above traditional search results, and AI Mode, a conversational feature that lets users chat back and forth with the search engine. Rather than matching searches to a list of external links, the new features use generative AI to answer complex questions directly on the search page -- a format critics say presents the automated summaries as if they are definitive answers.

PCWorld reports that Apple's new AI-enhanced Siri, now in public beta, takes a revolutionary business-like approach focused on practical tasks rather than companion-style interaction. The updated Siri efficiently organizes emails, retrieves specific information, and brainstorms ideas without attempting to bond with users like ChatGPT or Claude. This shift toward utility-focused AI assistance will reach wider audiences with the fall release of iOS 27 and macOS 27. After a couple years of false starts, the new AI-enhanced Siri has finally arrived on Apple devices via public beta. Its most revolutionary feature may be what it doesn't do - namely, it doesn't try to be your buddy. I've been testing the new Siri (which goes into wide release this fall once iOS 27, macOS 27, and the Apple's other OS 27's land) on my iPhone for the past few days, and I'm struck by how business-like it is.

Netflix says it's already used AI in'roughly 300' titles this year Netflix says it's already used AI in'roughly 300' titles this year Don't expect that number to shrink any time soon. Netflix hasn't made any secret of its interest in artificial intelligence, and now we have a sense of how those tools are being used in its content. In 2026, GenAI workflows have been used in roughly 300 of our titles, with the largest concentration of work in post-production, according to the shareholder letter detailing its second-quarter financials. The company named three projects -- (India), (Brazil) and (US) -- that used generative AI to create highly complex sequences, but the tech is becoming more widespread at this point. We already knew that Netflix had applied generative AI in at least one original show as of last July, but between making acquisitions and launching new specialized studios, its ambitions clearly extended further.

Child welfare referred Black parents to law enforcement 14,000 times in 7 years, despite no findings of abuse or neglect. After Robinson had a false positive test for cocaine that year, the hospital barred her from breastfeeding her daughter, and child welfare officials placed her under spousal supervision. Get your news from a source that's not owned and controlled by oligarchs. The mothers of Black newborns are more likely than those of White newborns to be referred to law enforcement over allegations of substance use during pregnancy, according to The Marshall Project's new analysis of child welfare data from eight states. The referrals are often the result of unreliable hospital drug tests performed at childbirth that are easily misinterpreted and produce false positive results as much as half of the time, which can prompt incorrect reporting to child welfare and law enforcement authorities.