
Introducing the Multimodal Knowledge Agent benchmark
We’re launching Multimodal Knowledge Agent, a new benchmark to evaluate how agents reason across visually-complex multimodal tasks.

We’re launching Multimodal Knowledge Agent, a new benchmark to evaluate how agents reason across visually-complex multimodal tasks.
Google is rolling out its latest AI model, Gemini 4 Carbon, to the general public after releasing it to a select group of "cybersecurity defenders" two weeks ago. Google claims Carbon is its "strongest model" yet, and capable of competing with the very best systems from Anthropic and OpenAI. Google says Gemini 4 Carbon "pushes
Self-interested intelligent agents that negotiate with humans should interpret nonverbal expressions and adapt their behavior accordingly. However, generic facial-affect models degrade across individuals and contexts, creating a generalization gap. We introduce an affect-adaptive negotiation strategy that incorporates Continual Learning with Imagination for Facial Expression Recognition (CLIFER) to personalize valence-arousal perception in real time. Differential feedback…
Users can now interact with Meta’s AI assistant through voice calls, text messages, or dictation on their Apple Watch.
The cybersecurity startup deploys a dedicated guardian agent for every AI agent in an organization’s environment.
The Russian delegation at the trilateral talks in Abu Dhabi has five members and six experts. Four of them are taking part in these talks for the first time, the independent Russian investigative outlet Agentstvo said.
Intezer researchers used AI agents to rebuild source code for Slack, Apple’s Finder and a TinyTurla-NG sample, from a backdoor family Cisco Talos attributes to the Turla group. They worked from the shipped programs and vendor documentation. The rebuilt Slack ran well enough that the team added a new feature to it: AI text autocomplete, powered by a small local model. Software vendors that count on compiling their programs to keep source code private are … More →
Foundation models such as GPT and Claude have shown strong capabilities in spatial understanding and embodied intelligence. These advances make agentic robotics a promising path toward artificial general intelligence in robotics. However, we find that using foundation models to control robots poses serious safety risks, and comprehensive safety evaluation is still lacking. We introduce RoboRisk, a comprehensive benchmark for evaluating safety in agentic robotics. Its 92…
Adobe is taking on Figma with a new prototype builder tool in Firefly.

Renaming files and resizing one graphic six ways are among my least favorite tasks in a creative project. Well, it looks like Adobe can now handle it through its AI agent. Adobe is expanding its Creative Agent. It’s basically an AI helper that carries out multi-step creative tasks based on plain-English instructions. More importantly, it
Desktop software company KORA Systems Inc. today unveiled KORA, a Linux-based operating system that builds artificial intelligence models and agents into the system itself. Qualcomm Technologies Inc. has spent close to a year working with the company to bring it to Snapdragon-powered PCs. The launch goes after what KORA sees as a ceiling on AI
Cyber insurance provider Cowbell Cyber Inc. today unveiled Cowbell AI Liability Insurance, a standalone policy for U.S. small and midsized businesses that covers losses caused by the artificial intelligence agents they deploy. The new policy is meant to deal with a problem that arises when businesses let agents move money or change customer records with
Adobe Inc. today announced an expansion of its CX Enterprise Coworker agentic artificial intelligence platform into its GenStudio content supply chain platform. The company said the integration move allows creative professionals to bring AI agents into production workflows to develop, manage and deliver personalized marketing content at large scale. The announcement, made at its Adobe
When it comes to matters of national security, it’s vitally important to ensure that the right intelligence gets to the right decision-makers, but doing this in today’s complex information technology environments is challenging, to say the least. Enter the artificial intelligence startup Lumbra Inc., which has just raised $25 million in early-stage A funding to
Adobe is committing itself to the notion that chatbots and AI agents could be your next design assistant, or even automate creative workflows and repetitive tasks on your behalf. Some of the tools and features announced today at Adobe's Max conference in Miami, Florida, include allowing your preferred third-party AI agents to access and change
Kora is set to launch as a free-to-use alternative to Windows, targeting users who rely heavily on AI models and agents.

Google is developing Swarm for Antigravity, a bee-marked command that coordinates multiple agents through a shared board and visible dashboard.
CAGE Challenge 4 is the benchmark on which learned multi-agent cyber defenders are compared. We reproduce its two reference defenders, KEEP and H-MARL Expert, in a pinned container and show that three things read from it measure less than they are taken to measure. First, the simulator silently discards any action asserted for an agent whose previous action is still running, and charges one point for a discarded Restore as for one that runs. The documented observation does…
Agent skills have emerged as an effective means of enabling LLM agents to solve complex, interactive tasks by encoding reusable knowledge without modifying model parameters. Recent approaches further improve skill adaptation by distilling reusable experience from agent execution trajectories and evolving skills based on downstream task feedback. However, existing methods primarily represent skills as natural language documents, intertwining procedural logic with descriptive…
Long-horizon personalized agents maintain an evolving memory by continuously deciding what to add, modify, or discard. However, learning an optimal update policy remains challenging. Conventional training relies solely on terminal question-answering accuracy, yielding a coarse feedback signal that treats all intermediate updates identically. Furthermore, update utility is inherently user-specific, as identical memory actions may yield opposite effects across different user…
Language-model agents increasingly tackle long-horizon tasks whose interaction histories exceed the model's active context. They must therefore decide what to write down and what to read back, and when to do either. Recent work has begun to use on-policy distillation (OPD) to make memory control part of the policy, relying on memory tools and compaction protocols that lie outside either model's pre-training. However, this weakens the distillation on both sides: the teacher…
Jensen Huang, the chief executive and co-founder of chipmaking company Nvidia, recently admitted to an extramarital affair. However, it’s not just a tabloid story: the woman involved, Alice Huang, has been accused of being a Chinese intelligence agent. Computer chips – specifically the chips used to produce and run artificial intelligence (AI) systems – have
Taxonomies are structured knowledge representations that support semantic search, recommendation, product catalog organization, and retrieval-augmented generation. Their automated construction from a fixed set of candidate entities requires more than plausible text generation: the resulting hierarchy must be complete, structurally valid, and semantically coherent. Large Language Models (LLMs) are promising for this task, but iterative free-form generation remains vulnerable…
AlphaProof Code agents wrote three million lines of proofs in Lean to verify Signal’s core encryption.
Good morning. Today I’m trying desperately not to buy Sony’s new nostalgia-bait headphones, reeling from how emotionally invested I was in Spider-Man: Brand New Day (yes, I know I’m late), trying to figure out what a “private, agentic notebook” is, and slowly coming around to the idea of the New York Times owning Letterboxd. Inbox is the tech newsletter that goes beyond the news cycle. We cover the most important 9to5Google stories from the last couple of days along with a…
Post-quantum handshake and Triple Ratchet get a formal security check
AI coding harnesses couple language models to shells, filesystems, and network sockets, turning the harness into an automated Living-off-the-Land (LotL) platform: once an adversary injects instructions into retrieved content, dual-use binaries such as bash, curl, and find become the payload. We argue that security in agentic systems is a structural property of the cognitive architecture and permission composition rather than an emergent property of individual models. Static…
Practical Web applications increasingly require agents to interpret long conversations or documents. For inputs exceeding the context windows of large language models (LLMs), memory agents process segments sequentially and retain past information in bounded memory. However, they typically divide inputs into flat token segments without preserving structure or hierarchy, causing three limitations for structured data such as API responses or HTML documents. First, fixed-length…
The sudden emergence of Nvidia CEO Huang's infidelity has created concerns about whether China has been able to access the cutting-edge chips designed by Nvidia that are powering the development of artificial intelligence
OpenAI said Control Room will connect every research inference workload, including agentic sampling for model training and evaluation, data processing, and more. It will work together with its misalignment monitoring systems.