AI news

The Alignment Research Center (ARC) and AIcrowd have launched the White-Box Estimation Challenge. The competition invites developers to improve estimation algorithms for random MLPs. A warm-up round is now open, with a total prize pool of at least $100,000 available in later rounds.

A new Datadog report finds nearly 70% of companies now use three or more AI models, a significant shift towards multi-model strategies. This approach allows teams to select the best model for specific tasks, optimizing for factors like cost, latency, and operational risk across different workloads.

A new report by STT GDC and Ecosystm finds that while 90% of Asian companies have started adopting AI, 71% are stuck in the early stages. Only 1% have reached a "Leader" level, highlighting a significant gap between initial adoption and true AI maturity.

Microsoft has released two open-source tools, Rampart and Clarity, to help developers secure AI agents. The tools are designed for safety verification during early development, addressing risks as AI agents gain more operational authority. This is part of Microsoft's push for continuous AI safety engineering.

Despite a surge in AI investments, many enterprise projects are failing to deliver expected results. According to Gartner, nearly half of companies struggle to show business value. Experts suggest the primary cause is not technical failure but unrealistic expectations set at the project's inception.

AI-powered tools are enabling non-technical staff in departments like HR and marketing to generate code, a trend called 'vibe coding.' This shift is democratizing software development, helping reduce backlogs and solve business problems faster, but it also introduces new risks that require IT oversight.

Google Cloud has announced the general availability of its managed Remote MCP Server for AlloyDB. This new service provides a direct and secure connection for AI models and agents to access real-time data stored in AlloyDB databases, improving the quality of context for AI-powered applications.

An engineer outlines a practical blueprint for building a secure and efficient AI-as-a-Service platform in a private cloud. The guide covers maximizing GPU usage, managing workloads with Valkey, securing LLMs against OWASP risks, and scaling data pipelines for enterprise use.

A new analysis highlights how advanced AI models can now autonomously discover software vulnerabilities at an unprecedented scale. This fundamentally changes the balance between offense and defense in cybersecurity, making traditional disclosure timelines potentially obsolete and requiring an urgent rethinking of security strategies.

A developer has released an open-source AI racing harness for AI Grand Prix contestants. Built with Rust and the Bevy engine, the tool provides a real-time flight software simulation that matches competition constraints, allowing teams to test their code while waiting for the official qualifier.

Gartner predicts that by 2025, 40% of enterprise autonomous AI projects will be partially derailed by governance failures. These gaps are often discovered only after production incidents occur, stemming from a flawed, all-or-nothing approach to trust and control, according to the analyst firm.

Salesforce has launched Headless 360, a new offering marking a significant strategic shift. The move is a response to the rise of AI agents and automated workflows, which reduce the need for users to stay within a single application. This change will impact IT architecture and budgeting.

Strava is now requiring developers to pay a subscription to access its API. The fitness platform cites the rise of zero-code AI tools and data scraping for degrading performance. This change marks a significant shift for developers building apps on Strava's data, reflecting a broader industry trend.

Former Google CEO Eric Schmidt was booed by University of Arizona graduates during his commencement speech. He told the class their task was to shape AI, prompting a negative reaction. Schmidt acknowledged their fears about job loss and the future were "rational," highlighting growing public skepticism.

Traditional FinOps practices often recommend downsizing resources with low utilization. However, for certain AI workloads like secure machine learning, low GPU compute usage can be misleading. These tasks may be memory-bound, not compute-bound, making "underutilized" GPUs essential for performance and avoiding higher costs.

Salesforce has launched Headless 360, a major architectural shift away from traditional application interfaces. The move allows AI agents, bots, and external tools to access Salesforce data directly via APIs, paving the way for new automated workflows and potential usage-based pricing models.

Payroll provider Remote has surpassed $300 million in annual recurring revenue and achieved positive cash flow. The company attributes this success to AI adoption, which led to a 50% increase in revenue per employee without expanding its workforce, highlighting AI's impact on operational efficiency.

In a recent discussion, experts from Dataiku and 1Password explored the next frontier of AI challenges. They covered the essentials of data governance, managing complex data supply chains, and the critical need for robust security frameworks to protect increasingly autonomous and interconnected AI agent swarms.

CIOs are under intense pressure to deliver measurable returns on AI investments. A new survey reveals a significant challenge: only 19% of AI projects are meeting their intended goals. This is forcing a shift from broad experimentation to prioritizing and scaling solutions that demonstrate clear business value.

Box founder Aaron Levie warns of "AI psychosis," where leaders overestimate AI's ability to replace jobs they don't fully understand. This mindset is driving workforce reductions, as seen with recent layoffs at companies like ClickUp, which replaced a significant portion of its staff with AI agents.

ClickHouse has announced major updates to its observability platform, ClickStack. The new releases include ClickStack Cloud in private preview, AI-powered Notebooks in beta, and a new MCP server. These changes aim to simplify setup, improve investigation, and enhance the platform's composability for developers and IT teams.

Microsoft Research has released Data Formulator 0.7, an open-source, AI-powered system for enterprise data analytics. The tool combines data connectivity, AI agent-guided exploration, and visualization in a shared workspace. A new Data Connectors feature simplifies integration with various data sources like databases and BI systems.

Vercel has integrated the MiniMax M3 model into its AI Gateway. This is MiniMax's first model with a 1-million-token context window and native multimodal capabilities, designed for complex tasks like software engineering, agentic web browsing, and multi-turn collaboration for developers using the platform.

Discussions about autonomous weapons, once confined to hypothetical scenarios at UN forums, are now confronting a new reality. The use of AI in modern conflicts is accelerating, raising urgent questions about ethics, international law, and the future of warfare for global leaders and the tech industry.

Mallika Rao, with experience from Twitter and Netflix, presents a new framework for evaluating production AI systems. She argues traditional metrics are outdated and introduces a five-layer evaluation stack and a maturity model to prevent silent failures and manage "evaluation debt."

GitHub reduced token consumption in its AI-powered CI workflows by up to 62%. The company achieved this by removing unused tools, replacing API calls with its CLI, and deploying daily automated agents to audit and optimize usage, offering a model for others to follow.

Microsoft has updated Azure Logic Apps with sandboxed code interpreters. This allows AI agents within workflows to safely generate and execute Python, JavaScript, C#, and PowerShell code, positioning Logic Apps as a platform for building AI-powered integrations.

The Linux Foundation has proposed an open standard for AI agents to discover and communicate with each other. The proposal suggests extending the existing Domain Name System (DNS) to create a universal, decentralized directory, avoiding the need for new proprietary registries and leveraging proven internet infrastructure.

DeepSeek has introduced reasonix, a new native AI coding agent. The tool is designed for high performance with features like advanced caching, aiming to provide a low-cost solution for developers. The announcement has generated significant discussion, highlighting interest in new developer tools.

Database company ClickHouse shared its year-long experience using AI coding agents. The team developed a practical framework to determine when agents are genuinely useful versus when traditional coding is better, moving beyond the general hype to offer specific, real-world guidance for engineering teams.

Payroll startup Remote has become cash-flow positive after reaching $300 million in annual recurring revenue. The company attributes this success to a 50% increase in revenue per employee, achieved through the strategic adoption of artificial intelligence tools without expanding its workforce.

Hugging Face has released an open-source project for 3D-printable humanoid robot legs. The hardware and software stack costs around $2,500 to build, significantly lowering the barrier for researchers and developers to experiment with physical AI and robotics software in real-world environments.

The Cannes Film Festival is embracing artificial intelligence, highlighted by a new multi-year partnership with Meta. The event showcased how AI is reshaping cinema, moving from a feared concept to a practical tool the industry is actively engaging with for creative and business applications.

AWS has launched a new version of its Resilience Hub, a service for managing application availability. The update introduces generative AI to analyze potential system failures, along with new tools for dependency discovery, modular policies, and organization-wide reporting to help teams improve system reliability.

AI agent frameworks like CrewAI and AutoGen are moving from demos to production environments for tasks like incident response. This shift is creating a critical new challenge: a lack of established tools and practices for monitoring and observing these complex, multi-step AI systems in real-world applications.

Microsoft is highlighting the security challenges organizations face when adopting AI. As companies like St. Luke’s and ManpowerGroup move to AI-powered models, establishing a strong security foundation for cloud, data, and identity becomes essential for enabling responsible innovation and managing new risks.

A journalist reportedly used publicly available facial recognition software to locate a long-sought fugitive. The individual, a member of Germany's Red Army Faction, had been hiding for decades before being identified and arrested, raising questions about the power and accessibility of AI tools.

A recent analysis reveals that leading AI models from major providers frequently disagree on basic, real-world facts. This challenges the assumption of factual consistency among frontier LLMs and highlights a fundamental reliability issue for developers and businesses building on this technology.

Snyk has launched Evo Continuous Offensive Security, a new AI-powered product for penetration testing. It aims to continuously find and fix vulnerabilities in applications, specifically addressing the security challenges posed by the rapid development and deployment of AI-generated code.

A new AI-assisted tool, highlighted by the CNCF, helps engineers migrate from ingress-nginx to the Higress gateway. The tool demonstrated its efficiency by converting 60 complex resources in about 30 minutes, significantly simplifying a common task for Kubernetes infrastructure teams.

OpenAI has launched Rosalind Biodefense, a new initiative providing vetted developers and U.S. government partners with access to a specialized AI model. The goal is to use frontier AI to advance biodefense, public health research, and pandemic preparedness in a controlled, secure environment.

Google DeepMind CEO Demis Hassabis stated that progress towards artificial general intelligence (AGI) is accelerating. He predicts AGI could arrive around 2030, but possibly as soon as 2029 or even earlier. This updated timeline suggests society has only a few years to prepare for its impact.

AI company MiniMax is teasing its upcoming M3 model, which features a new sparse attention mechanism. The company claims this innovation boosts long-context response speeds by up to 15.6 times. A technical paper detailing the new mechanism has also been released for developers and researchers.

YouTube is updating its policy on AI-generated content. The platform will now automatically apply labels to videos that use significant photorealistic AI, rather than relying only on creators to disclose it. These new labels will also be made more prominent to viewers for greater transparency.

A new startup, Human Archive, is paying gig workers in India to collect real-world data for AI and robotics. Workers wear camera-equipped caps and sensors, capturing physical interactions to create training datasets for embodied AI, addressing a critical data collection bottleneck for research labs.

Microsoft has launched MDASH, a new AI-powered security platform for large-scale vulnerability discovery. The system uses more than 100 specialized AI agents that work together to automatically scan, validate, and prove security flaws across complex codebases like Windows and other Microsoft software.

Google has updated its open-source AI framework, Genkit, with a new middleware architecture. This feature provides a programmable layer for developers to intercept and control model calls and tool execution, enhancing reliability, safety, and orchestration for production-level AI applications.

Anthropic's Project Glasswing, an AI-powered security initiative, has discovered over 10,000 high- or critical-severity vulnerabilities in widely used software. The project, which launched last month with around 50 partners, demonstrates AI's growing capability in automated vulnerability detection for critical systems.

AWS is enhancing its AI-powered service, AWS Transform, which helps modernize legacy applications. A new feature, AWS Transform custom, now allows organizations to create their own rules to automate code upgrades, framework migrations, and performance optimizations at scale, tailored to their specific needs.

Employees are increasingly using unapproved AI tools, creating security risks. A new guide outlines a five-step approach for companies to manage this "shadow AI." The goal is to establish practical governance and security policies without hindering employee productivity or innovation, balancing control with enablement for teams.

AI presents both opportunities and significant risks for growing businesses. Cybercriminals are leveraging AI to launch more effective and targeted attacks. For example, AI-automated phishing has proven to be 4.5 times more effective than traditional methods, creating new challenges for business leaders balancing growth and security.

AI company Anthropic has acquired Stainless, a New York-based developer tools startup. As part of the deal, Stainless will wind down its hosted products. The startup's tools were used by major tech companies, including competitors like OpenAI, Google, and Cloudflare, to streamline development processes.

Researchers have developed a new attack called CrossMPI that targets multimodal AI models. It uses nearly invisible changes in images to manipulate how the AI interprets both visual and text inputs. This technique bypasses safety measures without altering the original text prompt, creating new security risks.

AI safety and research company Anthropic has acquired Stainless, a product design and engineering firm. The Stainless team, including its co-founders, will join Anthropic to help scale its enterprise offerings and accelerate the development of its AI products, including the Claude model family.

A recent analysis explores the economic consequences of mass layoffs driven by AI automation. The argument suggests that if companies fire their workforce, they also eliminate a significant portion of their consumer base, raising questions about who will purchase their services in a highly automated future.

Cloudflare tested Anthropic's security-focused AI model, Mythos Preview, on its own infrastructure. As part of Project Glasswing, they used the model to scan over fifty internal code repositories to identify potential vulnerabilities and understand how attackers might leverage similar advanced AI tools in the future.

Prominent AI researcher Andrej Karpathy, a founding member of OpenAI and former head of AI at Tesla, has announced he is joining Anthropic. This move places a key figure from OpenAI's early days at one of its main competitors, signaling a significant talent shift in the AI industry.

Google has announced Gemini Spark, a personal AI agent designed to operate 24/7, even when devices are off. It can draft emails, manage documents, and monitor inboxes, with future plans to handle purchases. This marks Google's push towards more autonomous AI assistants amid intense industry competition.

Google announced a major overhaul of its search box at I/O. The new search bar will be more than a query tool; it aims to be an all-in-one assistant. It will dynamically expand for longer queries and offer advanced AI-powered suggestions that go beyond simple autocomplete.

Google is integrating AI-generated summaries, called AI Overviews, directly into its main search results. This feature is now the default for users in the U.S., with a global rollout planned. The goal is to provide direct, synthesized answers for complex questions, fundamentally changing the traditional search experience.