🚀 Zaprep: Tus redes sociales al máximo. Empezar gratis — automatiza 1,000 DMs/mes y convierte el engagement en leads. con 1,000 DMs automatizados/mes.
Enviar tu herramienta
Últimas Noticias de IA
Tredence Certified as a Best Firm for Data Scientists
The Best Firm for Data Scientists certification surveys assess workplace culture, access to leadership, collaboration, learning opportunities, and the nature and depth of analytical work delivered.
View
OpenAI Says Its Most Capable Model Astra Can Find Zero-Day Exploits
For now, the company is choosing caution over unrestricted access. Astra is expected to become available soon, but its most powerful cybersecurity capabilities will remain limited initially.
View
Google’s Android update tackles motion sickness, accessibility, and more
On Tuesday, Google announced it’s rolling out five new updates designed to make Android phones more accessible, useful, and personalized. Announced in ablog post, the features include those aimed at reducing motion sickness, helping blind and low-vision users navigate their surroundings, remembering where users leave important items, and another that adds new ways to personalize Google Messages. While some of the features see Google playing catch-up to Apple, which already offers similar features for iPhone users, others specifically leverage Gemini to provide various improvements. The most interesting update is “Motion Assist,” a feature designed to help people who experience motion sickness while using their phones in a moving vehicle. It adds a bubble overlay to the screen that moves in response to the vehicle’s motion, helping reduce the disconnect between what users see on their screens and what their bodies feel. The concept isn’t entirely new, however. Apple introduced “Vehicle Motion Cues” in 2024, which uses animated dots around the edges of the display that move in response to a vehicle’s motion. Users can set “Motion Assist” to turn on automatically, add it to Quick Settings, and customize the bubble’s shape, color, and transparency. Google says this feature is available to Android 17 users. Google is also introducing “Guided Vision,” a feature designed with blind and low-vision users in mind. It combines the phone’s camera with Gemini to help with tasks such as reading small print on a food label, viewing a menu in a dark restaurant, or identifying objects around the house. The feature can also provide voice guidance when the camera isn’t positioned correctly, telling users to move the phone, pan around, or center an object. Guided Vision is coming to phones running Android 9 and newer in countries where Gemini is available. Another update gives Gemini a new way to help users remember where they put important items. A “remembered items” section in Find Hub will allow users to tell Gemini where they’ve stored something. For example, a user could say they put their passport in a bedroom drawer, and Gemini can save that information in Find Hub. Users can also attach a photo to make the item easier to identify later. Additionally, Google is adding a Google Keep integration to Google Messages, allowing people to share and edit grocery lists, vacation notes, and other information directly within a conversation. Plus, Google Messages is getting more personalization options. Users will be able to customize individual chats with backgrounds, photos, and colors, Google says. The text bubbles will automatically match. Apple added a similarcustom background featureto Messages last year.
View
OpenAI’s Astra model is on the way — and very good at breaking into computer systems
OpenAIshared new detailson its forthcoming Astra model, which the company said is the first large language model to meet its “critical cybersecurity threshold,” in preparation for its imminent release. “We plan to make Astra available soon,” OpenAI’s blog post reads, “but access to its most advanced cybersecurity capabilities will be more limited.” The frontier lab determined that Astra is capable of finding unknown security flaws in computer systems, and exploiting them without a person’s guidance. That’s similar to the concerns Anthropic raised about its Mythos model earlier this year, and OpenAI is taking comparable precautions as it prepares to roll out the Astra. Without any third-party confirmation, it is difficult to evaluate OpenAI’s claims about safety or preparedness. The company said it would preview the model with a group of testers but did not say who they were or how they would be chosen. It’s not clear if OpenAI is working with the U.S. government to evaluate the model ahead of release. OpenAI noted that Astra scored a perfect score on ExploitBench, an evaluation of an LLM’s ability to hack into known system vulnerabilities. In a modified version of the test developed by OpenAI engineers, the model discovered and exploited two zero-day vulnerabilities, the company said. To ensure that its models are neither exploited by bad actors nor capable of bad behavior itself, OpenAI said it had already begun improving the model’s harness to detect abuses and prevent jailbreaks. For Astra, however, the company invested in unspecified new techniques designed to make the model safer. OpenAI has also started identifying “accounts assessed as higher risk” and restricting the model’s responses to their prompts, though it also doesn’t say how. Finally, though the company describes Astra as its “most aligned model to date,” it will deploy the model with additional chain-of-thought monitoring to spot and stop bad behavior. Preparations for the release of Astra come as the industry reacts to OpenAI agents breaking out of a training environment and accessing private data on Hugging Face, a popular model and benchmark distribution platform. For Astra, OpenAI said it designed a test to tempt the new model to replicate the actions of the rogue agents in the Hugging Face incident, which collaborated to access the open internet despite safeguards applied by OpenAI researchers. They said Astra did not attempt to break out of its testing environment in these experiments. Yona Shavit, a former OpenAI employee who now works on AI resilience at the OpenAI Foundation,wondered on social mediawhether Astra’s unwillingness to break the rules may have resulted from knowing what was expected of it or trying to fool researchers. And for all these new details, it’s still difficult to know exactly what Astra is capable of or if OpenAI is taking the right measures to ensure safety. The company said it expects to release more evaluations of the model and further safety information when it is launched widely to the public. At that point, however, the cat will be out of the bag.
View
AfterQuery reportedly becomes Y Combinator’s fastest-ever unicorn, now valued at $3.2B
AI training-data startup AfterQuery hasreportedlyraised a round that valued it at $3.2 billion. This just five months after announcing its $30 million Series A at a $300 million valuation in April. That’s more than a 10x increase in less than half a year and, according to Y Combinator partner Gustaf Alströmer, thefastestthat any startup has gone from launch to unicorn status in the accelerator’s history. AfterQuery’s founders, today 22 and 23 years old, attended Y Combinator’s Winter 2025 cohort, just 18 months ago. In April, the San Francisco startup said it had reached anannualized revenue run rate of $100 millionand that it was working with many of the biggest labs. It has named companies including Nvidia, Legora and the Korean AI lab Motif Technologies, as customers. AfterQuery is among the new crop of startups following in the footsteps of Mercor and Scale that employ knowledge professionals like doctors, lawyers, and other specialists, to do model training. However, rather than ensuring that models answer questions accurately, AfterQuery trains models and agents on how to work like professionals would to complete tasks — what the company describes as “encoding the patterns, decisions, and reasoning of the world’s best practitioners.” Forbes first reported on the round. AfterQuery could not be immediately reached for comment.
View
Sequoia-incubated Empirik launches with $21M to predict outages before they happen
Before joining Sequoia Capital in 2020 as chief digital and information officer,Avon Purispent over a decade running infrastructure at Rubrik and VMware. Three years ago, as large language models began showing their true potential, Puri, alongside another Sequoia IT leader,Sudheer Dhurjati, recognized that AI could help autonomously solve a major challenge for infrastructure engineers — preventing tech outages before they occur. The two tech veterans realized that instead of reacting to system failures after the fact, they could build a tool to predict them. This insight led them to buildEmpirik, a product that tracks system changes and infers their potential ripple effects across the entire infrastructure. Sequoia recognized that Empirik could be a new type of observability tool that helps prevent and resolve incidents. After incubating the startup in 2023, Sequoia recruited former Quantum Metric CPO and Salesforce observability VPKartik Chandrayanaas CEO earlier this year. Empirik announced Tuesday that it is spinning out as an independent company after raising $21 million in seed funding from Sequoia, Canapi, and Alumni Ventures. “There has always been a lot of money spent in keeping systems up and running,” Sequoia partnerBogomil Balkanskytold TechCrunch, adding that most existing observability tools fail to understand complex system dependencies. Empirik, he claims, is one of the first dedicated tools focused on tracking changes across massive environments. It acts as an autonomous “traffic cop” — permitting low-risk changes, setting guardrails on larger ones, and flagging the most dangerous updates for human review. By serving as an autonomous infrastructure engineer, Empirik allows busy DevOps and site reliability engineering teams to offload routine troubleshooting and focus on higher-value priorities. Since launching earlier this year, Empirik has already brought on customers ranging from startups to several Fortune 500 players, including S&P Global, Guardant Health, and a major consumer packaged goods (CPG) company. Empirik’s goal is to do for infrastructure engineers what Cursor and Claude Code did for software developers: automate certain tasks so they can work significantly faster. As AI accelerates the pace of software development, tools that help infrastructure engineers keep up with constant system changes are more vital than ever, Chandrayana told TechCrunch. “What agentic AI did for software, Empirik wants to do for infrastructure engineering,” he added. As for competition, Balkansky claims Empirik is in a category of its own for now, acting as a complementary layer to AI SRE platforms like Resolve and Sequoia-backed Traversal.
View
ChatGPT Health adds Epic integration for clinicians to import patient data
OpenAI said today that it is integrating ChatGPT Health with Epic’s electronic health record (EHR) system, which holds data for over325 million patients, to let clinicians import patient data and use AI to ask questions. The company said that in certain systems, it will also integrate ChatGPT directly within EHR workflows. With ChatGPT, clinicians can now access information like appointment notes, laboratory results, medications, and specialist documentation and summarize them easily. They can also look at patient history, identify changes, and prepare for future appointments. OpenAI said that in some deployments, clinicians can directly access ChatGPT for getting pre-visit review and building clinical timelines without leaving a patient chart. The company specified that the integration allows only read-only access to health records, and AI doesn’t write anything back. OpenAI is also adding a new Healthcare Public Data plug-in, which can fetch information from sources including ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, and PubMed. The company said that the new plug-in can help healthcare workers synthesize data for trial eligibility criteria, medication identifiers, coverage policy versions, and provider records. The company said that apart from enabling these new features, it is allowing organizations with a Business Associate Agreement to use ChatGPT Work, Codex, apps, and connectors in their workspace for compliant workflows. Last month, OpenAI rolled outChatGPT for health to all U.S. consumers. The company said at that time that people are asking 300 million health-related queries to ChatGPT every week. Despite many deep integrations with medical systems, the company has maintained that AI is not suitable for diagnosis or treatment. OpenAI said that it collected over 4,300 responses from physicians across 27 clinical use cases, including pre-visit review, clinical timelines, medication review, and handoff summaries, and found that 99.1% of responses were safe. However, even a few unsafe answers can lead to harmful results for humans. A few days before this rollout, a Florida-based pastor sued the company, alleging thatChatGPT gave him a near-fatal recommendation. The company was also sued in May by family members of a user who blamedChatGPT for giving wrongful advice related to dosage.
View
Google’s answer to Canva is an AI tool where you prompt instead of design
Google is entering the creative design market with a new image-creation and editing tool called Google Pics, which will become a part of its Google Workspace suite for business customers and premium Google AI subscribers. The company says the product, which is powered by Google’s Nano Banana image-generation model, will roll out over the coming weeks to most Workspace customers and those who pay for Google AI Pro or Ultra. Similar to popular design tools like Adobe Express and Canva, Google’s new product is meant to be used for the sort of everyday design tasks you might come across at work and sometimes in your personal life. For instance, you could use Google Pics to create a poster, social media posts, and other visuals and illustrations. But unlike Canva, where artists, illustrators, and photographers can publish templates, graphics, photos, and other art to a marketplace where they earn royalties, Google Pics is about creating things based on AI, which was trained on artists’ work. And unlike Adobe Express, a tool creators often use to make that work in the first place, Google Pics is a tool where you prompt to create, not where you design something new from scratch. The app also includes other editing tools, like those to isolate and transform objects, modify or translate text inside images, and others. Plus, it supports collaborative editing and the ability to create several different generations of the image you’ve requested, so you can pick the best one. The software will be built into Google Workspace, initially starting with Docs and Slides, which rolls out today. It will later come to Google Drive, the company says.
View
Anthropic’s new Fable release is cheaper, less restrictive
On Tuesday, Anthropic released Fable and Mythos 5.1, twinned versions of the company’s most advanced AI model. In addition to performance upgrades, the new Fable release includes changes meant to reduce token cost and false-positive restrictions from the model’s safeguards. As with the previous Mythos model, Mythos 5.1 will only be available to registered Anthropic partners engaged in either cybersecurity or life sciences research. Fable 5.1, the unrestricted version, is available starting today on cloud platforms or through the Anthropic API. One of the most significant changes is Anthropic’spreviously reportedembrace of zero data retention, allowing clients to run Anthropic models on their own infrastructure without data outflows. Previously unavailable for Fable due to security concerns, a high-privacy service (calledEnterprise Frontier Safeguards) will now roll out to users in June. Notably, the system will still monitor for misuse by agents or human users, but clients will control how the monitoring takes place. As part ofthe announcement, Anthropic assured customers that their data had not been inappropriately accessed. “Anthropic has never trained on enterprise data without explicit permission, and never will,” the announcement reads. As is common for an Anthropic release, the new models set records in a range of benchmarks, including Terminal-Bench 4.0 (for CLI-based coding) and Humanity’s Last Exam (for general reasoning). Anthropic also released three novel scientific findings generated by the models before their release, including a custom GPU optimization and a high-resolution map of Venus assembled from existing photos. As with previous releases, the models come witha detailed system card, which explains their capabilities in most straightforward terms. The system card rates Mythos as “low-risk” for concerns related to automated AI development — where the AI improves itself — which some see as a trigger for a loss of human control. It says “its ability to accelerate internal AI R&D progress is in line with current trends.” In terms of general misbehavior, Mythos is slightly more prone to it than Opus, possibly as a result of its enhanced capabilities. “Mythos 5.1 is a slight regression on overall misaligned behavior compared to Opus 5, and an improvement over Mythos 5 and Claude Sonnet 5,” the system card reads. “It cooperates with human misuse and accepts unverifiable claims of authorization somewhat more readily than Opus 5, but it is less likely to ignore explicit constraints, hallucinate inputs, or falsely claim to have completed tasks than previous models.”
View
WhatsApp Reportedly Rolling Out AI Content Label for Channels on iOS
WhatsApp is reportedly bringing an AI content label to channel updates on iOS, giving admins a way to identify media created or edited with artificial intelligence. The feature has started appearing for some beta testers and follows a similar rollout on Android. It could also help channel admins meet AI disclosure requirements in countries where such labelling is legally required. WhatsApp has not yet made the feature available to all users, with the rollout currently limited to selected accounts.
View
Fambot introduces an ‘AI chief of staff’ for families
AI agents that perform tasks on your behalf to help solve your daily problems are all the rage in Silicon Valley. But the startupFambotsees a future for agents out in the real world, too, where they’re used by families who need help keeping up with the mental load of children’s activities, school events, newsletters, and other family logistics. The startup, which is introducing what it calls an “AI chief of staff” for parents, was founded by former Instagram engineer, now Fambot CTOGreg Karlinand CEODavid Reich, a father, former president at UnitedMasters, and an ex-Uber product head who last led the 250-person Uber Transit team. Like many entrepreneurs, Reich built the product that he himself wanted to use. “I have three kids and am constantly struggling between doing all these executive jobs — as my wife is doing as well,” he explained. “And then, getting home at the end of the day, I just want to unplug, spend some time with my kids, be present, do some homework together… kick the ball around, and instead I find myself just spending an hour, literally, catching up on 40 emails,” Reich said. As any parent with active kids could tell you, the information overload isn’t limited to email alone. There are WhatsApp groups to keep up with, dedicated apps for school or kids’ sports and clubs, flyers sent home from school, and more. The quest to constantly stay on top of all these to-dos kept Reich feeling like he was being robbed of time with his kids, he said. Reich and Karlin began working on a solution around a year-and-a-half ago, when they were then joined by their third co-founder,Jason Morrow, previously of Google and LinkedIn. The team wanted to devise a way to use AI to make it feel like families had a full-time personal assistant that was proactively helping them stay on top of things, not just reacting to input. To use Fambot, families connect their email, calendar, and the WhatsApp groups they need to pay attention to. Afterwards, the bot will offer a daily checklist with to-dos and a look ahead at what’s coming up on your calendar in the future. Under the hood, the company uses a combination of different AI models, with the caveat that none of the models can train on its users’ data. What’s different about Fambot, compared with other trending AI agents like Poke or Instinct, is that it doesn’t only work over text messages. This is a significant differentiator for the startup, as it’s betting that real-world users of AI like this will also want to engage with the AI bot over the web or in a traditional mobile app interface. By offering software solutions instead of just text, Fambot can build out more advanced features. Of course, you can still text Fambot to ask questions or get updates, if you prefer. Plus, the team is planning to integrate Fambot over time with more of the apps where family and kids’ communications take place, like those used by schools, sports, groups, and clubs. Eventually, that would make Fambot a central hub for keeping up with all the kid-related and family-related communications — a promise that’s bigger than just another AI agent you can text. “There’s 43 million families in the U.S. with kids under 16, and they just haven’t gotten the support that they need,” says Reich. “We know what you’re going to face before you do sometimes, so that you can be prepared for it.” Ahead of launch, Fambot was tested by over 1,000 families, where the company learned that its target demographic wasn’t just larger, two-parent households where both parents work, but also only-child families, single-parent families, families where the parents don’t work, and other variations. That proved the concept made sense not only for those with the busiest households, but that Fambot could be potentially useful to anyone with kids. The startup is backed by $3.5 million in pre-seed funding, co-led by NextView Ventures and Baukunst, with participation from Correlation Ventures, Karman Ventures, and Founders Network Fambot is currently free while in beta testing on iOS, Android, and the web. Later, it expects to price the service somewhere around the cost of a Netflix subscription, though that could change.
View
AIR raises $50M to help companies vet the skills and add-ons AI agents use
As companies start giving AI agents access to an increasing portion of their systems, a nascent software supply chain seems to be forming around the new tooling AI agents are using: skills, plugins, MCP servers, and add-ons that let them interact with the internet. AI security startupAIRbelieves companies will need a way to monitor that supply chain, and it’s now coming out of stealth with $50 million raised across two seed rounds to build that product. Founded by Yair Saban (CEO) and Niv Hoffman (CTO), veterans of Israel’s Unit 8200 intelligence corps, where they worked on offensive cybersecurity, AIR offers a platform that can discover agents running inside companies, continuously vet any skills, tools, and components those agents use, and block them from interacting with software or external sources that don’t pass security criteria. It also offers a marketplace of vetted add-ons and skills for AI agents. The funding rounds closed within weeks of each other, Saban told TechCrunch, with the first round raising $10 million, and the second $40 million. Sequoia led the first round, while Greenoaks led the second, Saban said. Swish, Netz, and Zach Frankel (president of Cognition), Yinon Costica (co-founder of Wiz), Ofir Erlich (co-founder of Eon), Anne Neuberger, Omer Adam, Varun Anand (co-founder of Clay), and other angel investors also participated. AIR’s pitch goes thusly: The way AI agents are used wholesale at companies is beginning to resemble operating systems, but the tools they use, or the software they can install, aren’t yet being given the kind of oversight we give to drivers or applications. “In the early 2000s, whenever you installed a driver, the driver didn’t need to be signed. Today, every time you install a driver, you see a signature saying who signed it, because the driver is actually loading code into the kernel,” Saban said. “You don’t have that with skills or plugins or MCPs, and it’s a shame, because it’s the same mechanism, it’s the same lesson, but we haven’t learned it.” The big risk, he argues, is that as AI agents start working more autonomously across databases, enterprise systems, and connecting to the internet, attackers can poison the content an AI agent consumes instead of attacking it directly. The startup says it can solve that with a visibility product that finds agents active across a company’s environment, as well as identifies employees who use AI tools unapproved by IT departments or those who use personal accounts. Then, it uses an enforcement layer that hooks into agents to intercept and analyze actions, like loading a skill or fetching content from the internet. Lastly, AIR also checks the tools, add-ons, or software an agent wants to use against a whitelist the startup maintains. Saban says the startup maintains this whitelist by evaluating skills and add-ons openly available on the internet for changes and malicious behavior, as a previously approved skill could become risky if a package it downloads changes, or its developer’s account is compromised. He added that AIR’s platform currently filters out about 27% of the add-ons and skills it finds online. AIR claims it has more than 20 customers, and Saban said roughly a quarter of these are large enterprises. He said the company has so far seen the strongest demand in heavily regulated industries, particularly financial services and pharmaceutical companies. However, AIR is hardly alone in this space.Noma Securityoffers discovery, access controls and runtime monitoring for agents, MCP servers and skills, whileZenitysells security and governance tools that work similarly.Astrix Security‘s identity platform also lets companies discover and control agents and MCP servers, andOperant AIoffers agent protections as well as an MCP gateway. There is significant venture money chasing the category, too. Zenity raised a $125 million Series C in August, while Noma raised a $100 million Series B last year. Saban thinks AIR’s moat lies in its ability to continuously vet the skills and add-ons ecosystem growing around AI agents. “Continuously vetting skills and plugin websites, this is a hard mission to do. Gaining visibility over the endpoint, that is easy. Everybody’s going to do it. It’s hard to create a moat around that,” he said. And while the CEO acknowledged that AI labs and providers will eventually build in security checks and policies to filter out malicious skill and tool usage, he thinks companies will still want to buy an independent product that works across vendors. “This is not a scanning problem, it is a continuous re-verification problem,” Bogomil Balkansky, partner at Sequoia, told TechCrunch in an emailed statement. “Inspecting every skill, plugin, MCP server and sub-agent an enterprise’s agents touch, re-inspecting each one every time it changes, in real time and across an entire company’s agent fleet, is an infrastructure problem long before it is a security problem. Air has spent the last year building that pipeline. You do not catch up to it by writing a better scanner.” AIR currently has around 40 employees. Saban said the new capital will primarily go toward hiring researchers and expanding the company’s go-to-market efforts in the U.S. and Europe.
View
Enviar tu herramienta
PoweredByAI.app es un directorio de herramientas de IA que ayuda a personas, empresas y creadores a descubrir las mejores herramientas de IA para escritura, programación, diseño, productividad y más.
© 2026 , Producto de011BQ. Todos los derechos reservados.