Latest AI News

After a long and gloomy winter, many people are looking forward to some summer travel. Google has some new tools to help you plan, but like most of what Google does now, the new features lean heavily on AI. And unusually, the most interesting of these additions is launching first on iOS. Google says that lots of people tend to take screenshots when they’re planning a trip. Instead of letting those images become lost in your camera roll, Google will let you

In the growing canon of AI security, the indirect prompt injection has emerged as the most powerful means for attackers to hack large language models such as OpenAI’s GPT-3 and GPT-4 or Microsoft’s Copilot. By exploiting a model’s inability to distinguish between, on the one hand, developer-defined prompts and, on the other, text in external content LLMs interact with, indirect prompt injections are remarkably effective at invoking harmful or otherwise uni

The Gemini 2.5 branch will eventually replace 2.0, which was only released in late 2024. It supports simulated reasoning, as all Google’s models will in the future. This approach to producing an output can avoid some of the common mistakes that AI models have made in the past. We’ve also been impressed with Gemini 2.5’s vibe, which has landed it at the top of the LMSYS Chatbot arena leaderboard. Google says Gemini 2.5 Pro (Experimental) is ready and waitin

US experts who work in artificial intelligence fields seem to have a much rosier outlook on AI than the rest of us. In a survey comparing views of a nationally representative sample (5,410) of the general public to a sample of 1,013 AI experts, the Pew Research Center found that “experts are far more positive and enthusiastic about AI than the public” and “far more likely than Americans overall to believe AI will have a very or somewhat positive impact on

Most of Google’s AI efforts thus far have involved adding generative features to existing products, but NotebookLM is different. Created by the Google Labs team, NotebookLM uses AI to analyze user-provided documents. Starting today, it will be even easier to use NotebookLM to explore topics, as Google has added a “Discover Sources” feature to let the app look up its own sources. Previously, to create a new notebook, you had to feed the AI documents, web li

As AI hype permeates the Internet, tech and business leaders are already looking toward the next step. AGI, or artificial general intelligence, refers to a machine with human-like intelligence and capabilities. If today’s AI systems are on a path to AGI, we will need new approaches to ensure such a machine doesn’t work against human interests. Unfortunately, we don’t have anything as elegant as Isaac Asimov’s Three Laws of Robotics. Researchers at Google D

When Francesca Mani was 14 years old, boys at her New Jersey high school used nudify apps to target her and other girls. At the time, adults did not seem to take the harassment seriously, telling her to move on after she demanded more severe consequences than just a single boy’s one or two-day suspension. Mani refused to take adults’ advice, going over their heads to lawmakers who were more sensitive to her demands. And now, she’s won her fight to criminal

Google was caught flat-footed by the sudden skyrocketing interest in generative AI despite its role in developing the underlying technology. This prompted the company to refocus its considerable resources on catching up to OpenAI. Since then, we’ve seen the detail-flubbing Bard and numerous versions of the multimodal Gemini models. While Gemini has struggled to make progress in benchmarks and user experience, that could be changing with the new 2.5 Pro (Ex

Google started cramming AI features into search in 2024, but last month marked an escalation. With the release of AI Mode, Google previewed a future in which searching the web does not return a list of 10 blue links. Google says it’s getting positive feedback on AI Mode from users, so it’s forging ahead by adding multimodal functionality to its robotic results. AI Mode relies on a custom version of the Gemini large language model (LLM) to produce results.

Later today, a former Facebook employee, Sarah Wynn-Williams, will testify to Congress that Meta executives “repeatedly” sought to “undermine US national security and betray American values” in “secret” efforts to “win favor with Beijing and build an $18 billion dollar business in China.” In her prepared remarks, which will be delivered at a Senate subcommittee on crime and counterterrorism hearing this afternoon, Wynn-Williams accused Meta of working “han

Gemini models have been dropping so quickly, it can be hard to grasp Google’s intended lineup. Things are becoming clearer now that the company is beginning to move its products to the new branch. At the Google Cloud Next conference, it has announced initial availability of Gemini 2.5 Flash. This model is based on the same code as Gemini 2.5 Pro, but it’s faster and cheaper to run. You won’t see Gemini 2.5 Flash in the Gemini app just yet—it’s starting out

Yesterday, OpenAI counter-sued Elon Musk, alleging that Musk’s “sham” bid to buy OpenAI was intentionally timed to maximally disrupt and potentially even frighten off investments from honest bidders. Slamming Musk for attempting to become an “AGI dictator,” OpenAI said that if Musk’s allegedly “relentless” yearslong campaign of “harassment” isn’t stopped, Musk could end up taking over OpenAI and tanking its revenue the same way he did with Twitter. In its

Google has been finding ways to work generative AI into everything else it does, so why not its collaboration with the Wild Dolphin Project (WDP)? This group has been studying dolphins since 1985 using a non-invasive approach to track a specific community of Atlantic spotted dolphins. The WDP creates video and audio recordings of dolphins, along with correlating notes on their behaviors. One of the WDP’s main goals is to analyze the way dolphins vocalize a

Google may have finally found an application of large language models (LLMs) that even AI skeptics can get behind. The company just released its 2024 Ads Safety report, confirming that it used a collection of newly upgraded AI models to scan for bad ads. The result is a huge increase in suspended spammer and scammer accounts, with fewer malicious ads in front of your eyeballs. While stressing that it was not asleep at the switch in past years, Google repor

A gaggle of Gemini The model dropdown in the Gemini app is a bit convoluted, particularly as we see products like Veo 2 and Personalization popping up there. Google has been releasing so many preview models and new ways of using Gemini that it can be hard to know which option to choose for a given task. In fairness, Google is far from the only major AI player with this problem. Tulsee Doshi is Google’s director of product management for Gemini, which means

In a Black Mirror-esque turn, some cash-strapped actors who didn’t fully understand the consequences are regretting selling their likenesses to be used in AI videos that they consider embarrassing, damaging, or harmful, AFP reported. Among them is a 29-year-old New York-based actor, Adam Coy, who licensed rights to his face and voice to a company called MCM for one year for $1,000 without thinking, “am I crossing a line by doing this?” His partner’s mother

As the global artificial intelligence (AI) race presses on amid a US-China trade war, the Taiwan Semiconductor Manufacturing Company (TSMC)—a $514 billion titan that manufactures most of the world’s AI chips—is warning that it may not be possible to keep its customers’ most advanced technology out of China’s hands. US export controls require chipmakers to monitor shipments and know their customers to restrict China’s access to AI chips. But in a recently p

This requires retraining. Workers who used to walk the perimeter now have to read SAP dashboards, manage automated tickets, and work with the robots. They have to trust the sensors, and management has to make sure operators know they can take manual control if something unexpected happens.Companies need to take the rollout slowly. Because syncing physical robots with enterprise software is complicated, large-scale rollouts should start as small, targeted p