Cognition: Cognition releases SWE-1.7, trained from Kimi K2.7 and available at 1,000 tokens/second, claiming it nears GPT-5.5 and Opus 4.8 on benchmarks at a lower cost  —  Today, we're launching SWE-1.7, the most capable model we've trained so far.  It reaches frontier-level intelligence at a much lower cost …

Copy trading has evolved from a niche feature into a common way to scale execution and … The Evolution of Copy TradingRead more The post The Evolution of Copy Trading appeared first on .

Discussion | Link

Discussion | Link

Getty Images A person wakes in the middle of the night, overwhelmed and needing someone to talk to. But instead of calling a loved one or booking a counselling session, they open ChatGPT. Around the world, artificial intelligence chatbots are becoming companions, coaches, sounding boards, and, for a rising number of people, unofficial therapists. Studies have found that many users turn to AI to discuss personal struggles, seek emotional support, reflect on their feelings, and better understand their mental health. The appeal is easy to understand. Chatbots don’t judge. Unlike stretched mental health services in countries such as New Zealand and Australia, they don’t keep people on lengthy waiting lists. But as AI tools become more involved in mental health, it is becoming increasingly important to understand where the technology can genuinely help – and where its limits lie. Can AI recognise depression? Today’s chatbots can seemingly do everything – from answering complex questions to offering relationship advice – all while sounding remarkably human and empathetic. With mental health specifically, research has shown that AI systems can provide helpful information, encourage self-reflection, and offer emotional support in some situations. Some studies even suggest that AI-based mental health tools can help reduce symptoms of anxiety and depression when carefully designed and used appropriately. AI is also beginning to show promise in helping people practise cognitive reframing by encouraging them to consider alternative ways of interpreting difficult situations. At the same time, researchers, clinicians and regulators have raised serious concerns. AI systems can generate inaccurate advice – sometimes agreeing with or reinforcing harmful beliefs instead of encouraging people to seek appropriate help – and miss signs of crisis. An AI system may sound understanding, but it cannot truly understand the person behind the conversation. Unlike mental health professionals, AI is not held to the same professional or regulatory standards if something goes wrong. More than just providing information, mental health care relies on trust, empathy, clinical judgement and human connection. All of this is why many experts see AI as a tool to support mental health care, rather than something that can or should replace it. So, where exactly might it have a useful role? We in the University of Auckland’s 2DN research group have been investigating one interesting application: spotting signs of depression earlier. Depression often affects how people communicate. Changes in speaking rate, pauses, tone of voice, word choice and emotional expression can provide clues about a person’s mental state. These are examples of what researchers call “digital biomarkers” – measurable patterns in our behaviour or physiology that can provide clues about our health. Researchers are also investigating many others, including facial expressions, sleep patterns and physical activity. Our work explores whether AI can learn to recognise patterns from both speech and text. Rather than diagnosing people or replacing clinicians, the goal is to develop tools that support screening and monitoring, helping flag people who may benefit from further assessment. This is similar to how wearable devices can detect unusual heart activity without replacing a cardiologist. Instead, they provide clinicians with another piece of information to help inform decisions. AI’s promise and pitfalls AI might support mental health care in many other ways. It has the potential to expand access to services, support underserved communities, identify problems earlier, help people better understand and manage their own mental wellbeing. It can also reduce barriers to seeking help – and even personalise therapies by adapting support to an individual’s needs when sufficient high-quality data are available. But with these opportunities come obvious challenges. Mental health data is among the most sensitive information a person can share. Privacy, security and informed consent must be carefully protected. AI systems can also inherit biases from the data used to train them, potentially affecting how well they work for different populations. There is also the risk of over-reliance. Recent research suggests that people may place too much trust in AI systems, even when the technology is wrong. Because AI often responds in ways that feel supportive or validating, users may accept its advice without questioning it or seeking professional help. In mental health settings, that trust can have serious consequences. Still, it is inevitable that AI’s role in mental health – as with all other areas of life – will only grow in coming years. Its greatest value may lie in helping people better understand their mental wellbeing and support clinicians to identify risks earlier. Technology can recognise patterns. People provide empathy, trust and clinical judgement. The future of mental health care may likely depend on combining the strengths of both. Frederick Sundram receives funding from Lottery Health Research, the Marsden Fund and the Health Research Council of New Zealand. Partha Roop receives funding from the Health Research Council of New Zealand and Google DeepMind for research on networking protocols. Dushanthi Madhushika Manamalage and Reza Shahamiri do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.

Daniel Wiessner / Reuters: Block agrees to pay $45M and offer live customer support for Cash App to settle claims by 46 US states that the company failed to protect users from fraud  —  Block (XYZ.N) has agreed to pay $45 million and offer live customer support for its mobile payment service, Cash App …

<p><strong><a href="https://bun.com/blog/bun-in-rust">Rewriting Bun in Rust</a></strong></p> Jarred Sumner has been promising this blog post (<a href="https://x.com/jarredsumner/status/2053063524826620129">since May 9th</a>) about his Zig to Rust rewrite of Bun for significantly longer than it took him to finish the rewrite.</p> <p>Honestly, it was worth the wait. This is a detailed description of an extremely sophisticated piece of agentic engineering, featuring dynamic workflows, trial runs, adversarial review and all sorts of other interesting tricks.</p> <p>Jarred spends the first half of the post praising Zig for getting Bun this far. Then we get to a core idea in the piece, emphasis mine:</p> <blockquote> <p>Our bugfix list felt bad and I was tired of going to sleep worrying about crashes in Bun. I don't blame Zig for that - other users of Zig don't have the bugs we had, and mixing GC with manually-managed memory is an uncommon enough thing for software to need that no language really designs for it. We wouldn't have gotten this far if not for Zig, and I'll always be grateful. <strong>Until very recently, programming language choice was a one-way decision for a project like Bun.</strong></p> </blockquote> <p>Everyone knows you should never stop the world and rewrite a large piece of software from the ground up. Joel Spolsky highlighted that in <a href="https://www.joelonsoftware.com/2000/04/06/things-you-should-never-do-part-i/">Things You Should Never Do, Part I</a> back in April 2000!</p> <p>Coding agents powered by today's frontier models change that equation.</p> <p>Why pick Rust? It all came down to those challenges with memory management:</p> <blockquote> <p>A large percentage of bugs from that list are use-after-free, double-free, and "forgot to free" in an error path. In safe Rust, these are compiler errors and RAII-like automatic cleanup with <code>Drop</code>.</p> </blockquote> <p>A crucial enabling factor for the rewrite was that the Bun test suite was written in TypeScript, which meant it could act as <a href="https://simonwillison.net/tags/conformance-suites/">a conformance suite</a>. This allowed an agent harness to automate much of the initial port from Bun to Rust, initially as an experiment to try out an earlier version of the model we now have access to as Mythos/Fable.</p> <blockquote> <p>At first, I didn't expect it to work. A few days in, a high % of the test suite started passing and I saw how much the new Rust code matched up with the original Zig codebase. My opinion went from "this is worth trying" to "I'm going to merge this". [...]</p> <p>For most of those 11 days (and after), I monitored workflows - manually reading the outputs to check for issues and bugs, and prompting Claude to edit the loop to fix things.</p> <p>How do you review a PR with +1 million lines added? How do you start to build the confidence needed to responsibly merge large quantities of LLM-authored code?</p> <p>A language-independent test suite with a million assertions, adversarial code review and when something does go wrong, fixing the process that generates the code instead of hand-fixing the code.</p> </blockquote> <p>The new implementation of Bun has been live in Claude Code for nearly a month now:</p> <blockquote> <p>Claude Code v2.1.181 (released June 17th) and later use the Rust port of Bun. Startup got 10% faster on Linux but otherwise, barely anyone noticed. Boring is good.</p> </blockquote> <p>A perk of working at Anthropic is that you don't have to pay for your tokens - handy when the estimated cost is $165,000!</p> <blockquote> <p>Pre-merge, this took 5.9 billion uncached input tokens, 690 million output tokens, and 72 billion cached input token reads — around $165,000 at API pricing.</p> </blockquote> <p>This whole thing is a fascinating case study in taking on wildly ambitious projects with the help of coordinated parallel agents. <p><small></small>Via <a href="https://news.ycombinator.com/item?id=48837877">Hacker News</a></small></p> <p>Tags: <a href="https://simonwillison.net/tags/ai">ai</a>, <a href="https://simonwillison.net/tags/rust">rust</a>, <a href="https://simonwillison.net/tags/zig">zig</a>, <a href="https://simonwillison.net/tags/generative-ai">generative-ai</a>, <a href="https://simonwillison.net/tags/llms">llms</a>, <a href="https://simonwillison.net/tags/ai-assisted-programming">ai-assisted-programming</a>, <a href="https://simonwillison.net/tags/anthropic">anthropic</a>, <a href="https://simonwillison.net/tags/bun">bun</a>, <a href="https://simonwillison.net/tags/conformance-suites">conformance-suites</a>, <a href="https://simonwillison.net/tags/agentic-engineering">agentic-engineering</a>, <a href="https://simonwillison.net/tags/claude-mythos-fable">claude-mythos-fable</a></p>

The animal escaped after apparently arriving inside a package at Meta's Bangkok office, injuring one employee before finally being caught.

<p><strong><a href="https://openai.com/index/introducing-gpt-live/">Introducing GPT‑Live</a></strong></p> OpenAI <em>finally</em> upgraded the model used by ChatGPT voice mode!</p> <p>I've had preview access for a few weeks in the iPhone app, and the new model is very impressive. It also has the ability to spin off harder tasks to GPT-5.5:</p> <blockquote> <p>For questions that require web search, deeper reasoning, or more complex work, it delegates to our latest frontier model behind the scenes and brings the result back into the conversation when it’s ready. While it works, GPT‑Live can keep talking with you and maintain the flow of conversation. At launch, GPT‑Live will use GPT‑5.5 in the background. As we release new frontier models, we’ll continuously update the model used by GPT‑Live.</p> </blockquote> <p>The previous voice mode in the ChatGPT app was based on a GPT-4o era model, with a knowledge cut-off some time in 2024. I had mostly stopped using voice mode because the age and relative weakness of the model greatly limited how useful it was as a brainstorming partner.</p> <p>During the preview period I encountered a pretty obscure bug: the model was interrupting me to laugh at things I said, which weren't even intended as jokes! It felt rude and condescending - I reported it to OpenAI and as far as I can tell they made some tweaks and it's now less likely to happen.</p> <p>From looking back at my transcripts I think it was this bit that triggered the interrupting laugh:</p> <blockquote> <p>so where are the owls when they're not, like before dusk? The owls exist, right? Are they hiding in holes? Where are they hiding?</p> </blockquote> <p>My longest conversation with the new model has been a full hour while walking the dog (and <a href="https://simonwillison.net/elsewhere/sighting/">taking photos of pelicans</a>). I have not yet managed to take a photo of an owl. <p><small></small>Via <a href="https://news.ycombinator.com/item?id=48834405">Hacker News</a></small></p> <p>Tags: <a href="https://simonwillison.net/tags/text-to-speech">text-to-speech</a>, <a href="https://simonwillison.net/tags/ai">ai</a>, <a href="https://simonwillison.net/tags/openai">openai</a>, <a href="https://simonwillison.net/tags/generative-ai">generative-ai</a>, <a href="https://simonwillison.net/tags/llms">llms</a>, <a href="https://simonwillison.net/tags/multi-modal-output">multi-modal-output</a>, <a href="https://simonwillison.net/tags/llm-release">llm-release</a>, <a href="https://simonwillison.net/tags/speech-to-text">speech-to-text</a></p>

Bloomberg: Sources: Reno-based AI chip startup Positron is in talks to raise ~$750M in two phases, at valuations of $3.5B in the first tranche and ~$5B in the second  —  AI chip startup Positron is in talks to raise about $750 million in financing during a round with two phases, part of a groundswell …

The saga of Musk's tussle with the SEC over how he disclosed his growing stake in Twitter (now X) has come to an end.

Meta’s new AI imaging model enables users to create competitive ad content on the social media giant’s platforms.

The $300 million round is expected to be led by Menlo Ventures, Sifted reported.

Boone Ashworth / Wired: The FTC settles with John Deere over a 2025 right-to-repair lawsuit, requiring the company to provide farmers and shops with equipment and software for 10 years  —  After more than a decade of pushback, farmers and repair advocates have won access to equipment and services John Deere had long kept under its control.

Discussion | Link

The National Highway Traffic Safety Administration said emergency scenes are not "edge cases."

AI cheating leads to "a failed society," professor says.

The Latin American nation's cybersecurity plan — still in the expansion phase — has to survive its own knockout round during the FIFA World Cup.

OpenAI: OpenAI says it found widespread task issues in SWE-Bench Pro, estimates ~30% of tasks are broken, and retracts its earlier recommendation to adopt the benchmark  —  Through a detailed audit, we find widespread task issues in SWE-Bench Pro and estimate that ~30% of the tasks are broken.

Lionel Messi and Cristiano Ronaldo are betting on AI, health tech, and startups. Mohamed Salah is taking a more traditional route beyond football.