In the spring of 2022, Google began reaching out to a series of writers, including award-winning science-fiction authors and New York Times bestsellers, for feedback on an experimental writing tool. Four years later, their participation has corners of the literary world up in arms. Google researchers knew at the time that they were approaching a breakthrough point with large language models (LLMs). DeepMind, the company's AI division, was internally testing a natural language generation model called LaMDA which had demonstrated emergent writing capabilities. That model would eventually form the basis for Google Gemini.

Way back in the summer of 2017, AI researchers at Google put out a paper called "Attention Is All You Need," in which they described a new type of neural network called a transformer. It proved to be very good at processing long sequences of data, especially text. Nine years on, transformers are the engines inside every major large language model on the market. "The entire AI industry is built on transformers," says Justin Dangel, cofounder and CEO of the AI startup Subquadratic. "They are one of the most important innovations in the history of computer science, and they've changed the world." But transformers are starting to show their age. Many of the recent advances in LLMs, such as the development of so-called reasoning models and their ability to handle large amounts of input at once, are not neat extensions of that core technology but workarounds that patch over some of its fundamental flaws. A growing number of scientists and engineers are now asking what's coming next. LLMs are not going anywhere, but the way they get built is up for grabs.

Will AI help you do your job or replace you? Artificial Intelligence (AI) companies are making vast claims about the ability of their tools to replace human labour. Some jobs will be automated, others will be augmented. The bosses of the world's biggest companies are diverting vast sums into these tools, partly with the knowledge that they could save money on headcount. Flat is the new up, we are told, in terms of the size of a company's workforce as investors ask whether jobs should be done by new recruits - or, instead, armies of AI Agents, virtual workers tasked with doing specific roles, some of them relatively skilled.

The next time you apply for a job, AI may screen your résumé before any human sees it. But there's good reason to question whether AI will judge you fairly. Researchers already know that LLMs pick up human biases from their training data. New research suggests that LLMs can also develop their own biases from experience--and stereotype job applicants more than humans do. As AI companies race to build agentic models that remember the tiniest details about users, they may be handing them ammunition for forming those biases.

"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.

The company says it has found a new window into how its models arrive at answers. We spoke with senior editor Will Douglas Heaven about it. Anthropic--currently the world's most valuable AI company, with a nearly $1 trillion valuation--has a reputation for publishing strange and heady research. It's looking into whether AI models can feel pain, for example, and will sometimes cut off chatbot conversations if it suspects users are "abusing" the model. One niche that Anthropic spends more time and money on than other AI companies is called mechanistic interpretability, which means looking inside the complex math of an AI model to learn why it comes up with one particular output and not another. It's complicated stuff; there are millions of data points that might contribute to any result, and wading through them can look more like word salad than anything useful.

The AI firm Anthropic has developed a technique that has given it the clearest glimpse yet at what's really going on inside large language models as they answer questions or carry out tasks. What they found ranges from the mundane to the unnerving. Researchers at the company built a tool called the Jacobian lens (or J-lens) and used it to uncover a hidden area, which they named the J-space, inside Claude Opus 4.6, a version of Anthropic's flagship LLM released in February.

Could the next great novel be written by AI (and would you even be able to tell)? Can you tell which, if any, were AI generated? "The hotel is in a great location for everything. Lots of places to eat and drink. The hotel itself is always abuzz. The tavern located on the ground floor is definitely a must. Food, service, prices and atmosphere were great." "A good hotel, though the room had the proportions of a well-appointed lift.

We deploy a simple into our everyday lives as search engines (Jin et al., 2025; statistical framework based on risk control (Angelopoulos Xiong et al., 2024), coding assistants (Zhao et al., 2023), et al., 2022) that converts any safety signal into a binary and companions (Zhang et al., 2025a). As their applicability grows, so does the potential harm caused by malicious decision rule, and offers statistical guarantees on the false LLM outputs. Despite remarkable performance across a alarm or missed detection rate. The framework is universally applicable to different monitoring purposes and can leverage wide range of tasks, LLMs remain prone to generating halarbitrary proxy signals. Through experiments on mathematlucinated, factually incorrect (Ravichander et al., 2025), or ical problem solving and red teaming conversations, we harmful output (Yu et al., 2025) when deployed.

Let's start with a game. Open up your chatbot of choice--Claude, ChatGPT, Gemini--and type "Give me a random number between 1 and 10." You're going to get 7. Almost always. Now type "Another" and you'll get 3 or 4. Type "Another" again and you'll get 8 or 9. That won't work every time--but if it did for you, you may wonder if I have superpowers.