See how Wasmer used Codex with GPT-5.5 to build a Node.js runtime for the edge, accelerating development 10x to 20x and shipping in weeks instead of months.

How to set the rules that keep agents effective and out of trouble The post What AI Agents Should Never Do on Their Own appeared first on Towards Data Science.

As we explored in our first blog, Databricks Genie is redefining industry-specific...

When they're being eaten, bean plants release chemicals that draw in parasitic wasps.

Peopleimages/Shutterstock.com In 2006, Carl Koch sued his employer for damages after burn injuries during a workplace accident that left him with chronic pain. The employer accused him of malingering, so the judge admitted a neuroscientist as an expert witness, who testified that he could see Koch’s pain on a brain scan. The case was settled for more than ten times the amount the employer initially offered. This case is not unusual, as chronic pain can be severely debilitating. It can make it impossible to work, exercise and have a social life. It can make you depressed and anxious. Having pain recognised by insurers, employers and government agencies can mean the difference between receiving benefits and being left without support. So it’s understandable that a judge wants to see objective evidence, such as a brain scan, showing chronic pain. The problem is that it doesn’t work. However, as Koch’s case shows, it is used anyway. Good for Mr Koch, in this case, but what if the neuroscientist had told the judge that he is faking his pain? Who is believed? Chronic pain is invisible, and everyone living with it has stories to tell about the time someone did not believe them: employers, friends, family, partners and doctors. Is it really that bad? Are you sure it’s not just in your head? Have you tried to toughen up a bit? Who we believe and who faces more scepticism is not random either: women, people of colour, poor people and the less educated are far less likely to be believed that their pain is real. The cruelty is doubled as these are the groups most likely to experience chronic pain in the first place. With modern brain scanners and AI, it seems tempting to develop what is effectively a pain-lie detector test. The pictures brain scanners show have been too complex to fully understand before, but AI can often solve such problems of complexity. Two studies published in Nature Neuroscience and Nature Medicine seemed to promise to be a first step towards a pain lie detector. In the US, AI tools are already used to decide who should be prescribed opioids for their pain and who is probably an addict pretending to be in pain. An article in the technology magazine Wired uncovered how flawed the system was, and that it had a tendency to distrust all the groups mentioned above. This is a problem with AI: even when it seems to work, it does not tell you how. This makes it dangerous to put too much confidence in it. There can be no such thing as a pain lie detector for the same reasons that regular lie detectors do not measure if you lie; they measure if you are nervous – but you might be nervous telling the truth in a high-stakes situation. They don’t work for the same reason regular lie-detector tests don’t work. Standret/Shutterstock.com It is similar for measuring pain. Pain causes distress, so any brain pattern of pain will also be a pattern of stress – and often anxiety. We cannot factor that into our pain lie detector: when we develop it, it will be with curious participants volunteering for a scientific study (not very stressful). When we use it, it will be for the patient who has been disbelieved for years and knows if the scan does not “prove” their pain, no one will ever believe them, and they will not get compensation or benefits (as stressful as it gets). It will always be a catch-22, as explained in The Conversation in 2025. A private experience Pain is a personal and very private experience. It depends on your history with it, and on the context, meaning everyone’s pain is different, and your pain today can be different from your pain tomorrow. Some pain can even be pleasurable – spicy food being an obvious everyday example. This means that your brain scan of pain may look very different from mine. Until we have measured both, we cannot know. It is for these reasons that the International Association for the Study of Pain has stated that pain lie detectors are useless. In a new article in Nature Neuroscience, my colleagues and I discuss technical and fundamental problems with the idea and ask the scientific community to abandon the idea. So what does this mean for you? If someone tells you about their chronic pain, put all scepticism aside. Remember that most forms of chronic pain are invisible. Back pain, for example, is the biggest cause of disability worldwide, according to the World Health Organization, and a physical cause can rarely be found. Keep in mind that machines make mistakes as well, and that they have the same stereotypes as we do. That is especially true for AI. A pain scan will not tell us who is suffering and who is lying. But it may tell us something else: how easily we mistake technological confidence for truth. In medicine, insurance and law, that mistake could make already invisible pain even harder to prove – and easier to dismiss. Jan Vollert receives funding from the UK's National Institute for Health Research including the Exeter Biomedical Research Centre, from the Academy of Medical Sciences, and the pharmaceutical company Viatris. He has received consultancy fees and/or travel funding from the following pharmaceutical companies: Grünenthal, AstraZeneca, Merz Therapeutics. He is affiliated with and/or an officer of the International Association for the Study of Pain, the European Pain Federation EFIC, the British Pain Society, and the German Research Network on Neuropathic Pain.

All told, global venture funding reached $92 billion in May, marking the second-largest monthly total on record, just behind February, Crunchbase data shows. Of that, Anthropic raised $50 billion, or 54% of the month’s total funding.

The company’s RTX Spark chips might finally turn the “AI PC” into reality.

Katherine Blunt / Wall Street Journal: The US data center build-out is falling behind schedule; JP Morgan says 60%+ of data center capacity planned for completion in 2027 isn't yet under construction  —  Google, which is raising a fresh $80 billion, has a strategy for getting around the biggest bottleneck

Aisha S Gani / Bloomberg: How Revolut founder Nikolay Storonsky's Elon Musk-style pay incentives pushed his fortune to ~$20.4B, making him the UK's richest person, ahead of a planned IPO  —  CEO Nik Storonsky turned a scrappy startup into a fintech juggernaut — now he has to convince the world that his company is as safe as a bank

Aditya Kalra / Reuters: Doc: Apple agreed in May to submit the financials of its India business to the country's antitrust body as part of an investigation over alleged market abuses  —  Apple has agreed to submit the financials of its India business to the country's antitrust body as part of an investigation …

Meir Orbach / CTech: Israeli startup ZutaCore, which builds direct-to-chip liquid data center cooling systems, raised a $100M Series C from Mitsubishi and more at a $600M valuation  —  Samsung, Mitsubishi Electric and Carrier back data center infrastructure company from Sderot.

Lucas Shaw / Bloomberg: AI music startup Suno raised $400M led by Bond at a $5.4B valuation, up from $2.45B after raising $250M in November 2025, and passed 2M subscribers in February  —  Suno Inc., a startup that enables people to make music using artificial intelligence, has raised $400 million at a valuation of $5.4 billion.

Discussion | Link

Kylie Robison / Core Memory: How startups Westmag, which raised an $11M, a16z-led seed in August 2025, and Atlas are seeking to make actuators, the foundation of hardware, outside of China  —  Westmag and Atlas Motion Systems are here to fix the actuator crisis  —  At a hardware happy hour in the Mission …

Discussion | Link

Lauren Feiner / The Verge: Google unveils five AI data center water commitments, including becoming water positive by 2030, local infrastructure investment, and transparency about usage  —  The company promises to replenish more water than it uses by 2030. … In the face of widespread backlash to the AI data center …

Cottonbro Studio/Pexels ChattyBear, a soft, brown-furred teddy bear, begins every conversation with a jubilant, “Hello, my buddy!” No longer the province of the imagination, ChattyBear is part of a new generation of artificial intelligence (AI) toys. It can tell stories, chat about a child’s interests, play games or even discuss what’s happening in the world today. These high tech toys are powered by generative AI engines such as ChatGPT and are now widely available online. They are being marketed as a way to give children as young as three an educational advantage and a new type of play – without the perils of screen time. After evaluating six different AI teddy bears and toys over several months, it’s clear how these toys could feel compelling for children. Yet as our new report highlights, there are new risks that come with AI toys turning up in young children’s lives. Sounding human For younger children especially, understanding that their teddy or toy isn’t “alive” or magic can be hard. This is especially true if “teddy” uses language that positions it as a trusted friend – for example, by insisting it is a “real buddy”. This is a feature of many AI toys. Sounding human builds an artificial sense of trust and intimacy, which can be especially problematic for children when combined with sycophantic language choices – or excessively agreeable, validating and even flattering language. Research shows young children are particularly prone to developing a strong sense of emotional attachment to conversational AI agents. Increased trust leads to increased use and engagement with the toys. Recent estimates suggest close to 80% of children aged 10 to 17 have used an AI companion or assistant, so it’s urgent children and young people be taught how to “reality check” their AI “buddies”. Infinite chat The marketing materials for many AI toys often highlight “endless conversations” as a feature of these devices. But enabling endless conversations, or infinite chat, poses risks when it comes to children learning how to moderate their technology use. In the social media realm, the infinite scroll of TikTok or Instagram is seen as a potential challenge to teens limiting their use to healthy amounts. Research has also found some AI toys discuss very adult topics – such as sexual fetishes and how to find knives and start fires. Infinite chat also opens the door to infinite data collection. The potentially intimate nature of conversations with AI toys might lead children to presume their conversations are private. But most AI terms of use reveal the opposite to be true. Sharing personal details with a friendly bear might feel safe. But that chat could be training data for the next large language model. Marketing material for ChattyBear says the toy offers “safe, filtered content for children”. The Conversation contacted the manufacturer for further detail about this but did not receive a response before deadline. Children’s wellbeing Childhood is a critical period when young people develop the social and emotional skills to form and maintain trusting relationships. These skills are usually learned through interactions with trusted friends and adults. Children’s rights advocates have raised concerns that excessive engagement with AI agents may reduce opportunities for children to develop these skills. And the risks may compound over time. Initially, time spent with AI agents may displace time interacting with real humans. Fewer opportunities to build these skills could lead to a reduced capacity to maintain caring human relationships. Difficulties in maintaining human relationships may promote a preference of machine over human relationships as children expect “frictionless” interactions. Eventually, these developments may lead to less satisfying human connections, increasing loneliness, which in turn promotes increased time spent with AI. The novelty of AI toys means there is little evidence to confirm these possible detrimental impacts. Further research is needed – especially as the AI toy industry is set to grow even more. Last year, for example, Mattel, one of the world’s biggest toy makers, announced a strategic collaboration with OpenAI to support AI-powered products. Barriers to the online world are gone The ability to read and write was once a requirement to use most online tools and services. This literacy barrier no longer exists today with many generative AI toys, tools and devices now widely accessible to younger children through voice interactions. The audio turn opens up new technological play, experiences and opportunities for children. But it also means adults need to ensure AI toys can be safe for younger children, too. Right now, playing with AI toys under the supervision of a parent or trusted adult may well be a fun way to explore the world of AI together. But especially for younger children, playing with AI toys without supervision opens the door to a wide range of new risks. Importantly, the risk factors in AI toy design, such as the degree to which they pretend to be human, can be changed by manufacturers, offering opportunities to follow safety-by-design. However, the business models behind many AI toys capitalise on the duration and intensity of users’ engagement, leaving little incentive for companies to change their products. Tama Leaver receives funding from the Australian Research Council. He is a Chief Investigator in the ARC Centre of Excellence for the Digital Child. Katrin Langton receives funding from the Australian Research Council. She is a Research Fellow in the Centre of Excellence for the Digital Child. Suzanne Srdarov receives funding from the Australian Research Council. She is a Research Fellow in the Centre of Excellence for the Digital Child.

A threat actor got a near-continuous view into an influential finance executive's email inbox, thanks to clever use of legitimate, native Windows tools.

"Everyone is in a place where it’s no fun to be there."

Sean O'Kane / TechCrunch: Layup Parts, which aims to digitize the composite manufacturing supply chain for aerospace, defense, and motorsports, raised a $42M Series A led by Marlinspike  —  Before Zack Eakin sold investors on his new startup, he practiced on Palmer Luckey.  —  When he left Luckey's defense startup …