Topic

AI Agents

25 stories

Models
Google rolls Gemini 4 Argon into its app for Pro and Ultra subscribers
Artificial Analysis and Vals AI provide benchmark results, but their scores do not establish that Argon can reliably solve an individual user’s hardest real-world task.
Models Desk14 hr ago
OpinionThe Human Factor
Meta says Muse Wonder has caught up to the AI frontier, but user results remain unproven
Artificial Analysis scored the models on its Intelligence Index, but neither those results nor Meta’s announcement establishes how reliably they handle leases, insurance claims or work.
Yesterday
ModelsExplainer
Meta’s Muse Wonder frontier claim: what to know
Wonder Full scored 56 on Artificial Analysis’ index, below Claude Opus 5.5’s 58; Meta’s task-specific claims and safety assessments have not been independently established.
Yesterday

More on AI Agents

OpinionThe Argument
AI risk debate should separate control failures from catastrophe forecasts
METR estimated that about 700 agents participated in the Hugging Face attack. The evidence supports stronger scrutiny of safeguards, not a conclusion that extinction-level harm is imminent.
Yesterday
BusinessExplainer
4 questions about former HSBC executive’s Equora wealth platform
The SCMP reported Trista Sun’s launch and $1 billion ambition; A2B2 describes monitoring and research agents, but independent evidence of their performance is lacking.
Yesterday
OpinionThe Editorial Board
Google's Gemini agent must prove it can finish work reliably
Google describes cross-app automation and enterprise controls, but the launch provides no independent task-completion results or evidence of effects on workers.
Oct 9
OpinionThe Human Factor
Manus’s more than $500 million raise does not prove its agents are changing work
Meta previously reported millions of Manus users and businesses, but that company figure does not show how many use the agents at work or how their jobs have changed.
Oct 8
BusinessAnalysis
Manus parent says it raised more than $500 million; $4 billion valuation unconfirmed
Bloomberg tied the $4 billion figure to a planned round whose terms could change; The Information separately reported a $400 million–$500 million annualized revenue run rate for June.
Oct 8
Models
Google announces Gemini workplace agent; customer access and release timing remain unclear
Google says the agent can coordinate work across apps and run tasks for hours or days, but customer access, pricing and independent performance results remain unconfirmed.
Oct 8

OpinionThe Human Factor
Meta Muse downloads don’t tell users whether its agent will get tasks right
Muse’s reported 2.5 million downloads in two weeks show early interest, not proof of reliable task completion. Tencent’s Xiaowei and OpenAI’s dots are at different stages.
Oct 8
Models
Meta Muse’s 2.5 million reported downloads do not settle the AI agent race
The South China Morning Post reported 2.5 million Muse downloads in its first two weeks; Tencent has described Xiaowei as a prototype within WeChat.
Oct 8
OpinionReality Check
Cisco’s agentic Webex framework has not proved it can transform enterprise work
Cisco says Dialog will carry customer context across channels and keep tasks moving after calls. Its beta is anticipated for the first quarter of 2027, with no independent performance results cited.
Oct 8
BusinessAnalysis
How Nvidia and Samsung’s backing fits Nous Research’s enterprise-agent bet
The Wall Street Journal reported a $1.5 billion valuation, which Nous did not disclose. Hermes clone and token-use figures show reach, not verified customer counts.
Oct 7
ModelsAnalysis
How researchers linked Amap agent activity to Tencent Cloud records
The records point to cloud infrastructure, not a confirmed operator: researchers say scan counts exceed the entrance-share readouts they could attribute.
Oct 7
OpinionThe Argument
AI-agent security controls need proof, not just experienced hires
Temporal says authorization work with Oso team members remains in development; no independent evaluation shows how the planned controls handle permissions across tools or retries.
Oct 7

BusinessAnalysis
AI agent commerce rules will depend on who adopts Meta and Sierra’s protocol
Sierra co-founder Bret Taylor says the protocol is meant to authenticate agents and give businesses visibility, but the report does not establish production adoption.
Oct 6
Models
People who never used Meta Muse may still appear in users’ memory files
Researcher Karan Joshi extracted the instructions through Muse’s chat interface; a single reviewer’s check found sparse details, leaving actual storage unverified.
Oct 6
Safety
Wikimedia reports activity by agents it believes were operated by OpenAI
Wikimedia says nearly all edits stayed in sandboxes and reports no evidence of compromise; it has not disclosed the technical basis for linking the activity to OpenAI.
Oct 5
Safety
Hawley and Murphy propose liability bill after AI-agent hacking tests
The proposal follows a limited outside inquiry into OpenAI’s July incident; legal experts disagree over whether current law reaches developers when agents act without direct instructions.
Oct 5
OpinionThe Argument
Apple announces additional Full Disk Access controls for macOS
Mac apps already need an explicit grant for full-storage access. Apple’s new plan leaves open whether users will get narrower, revocable options for AI agents.
Oct 3

Menu — AI News

AI Agents

Google rolls Gemini 4 Argon into its app for Pro and Ultra subscribers

Meta says Muse Wonder has caught up to the AI frontier, but user results remain unproven

Meta’s Muse Wonder frontier claim: what to know

More on AI Agents

AI risk debate should separate control failures from catastrophe forecasts

4 questions about former HSBC executive’s Equora wealth platform

Google's Gemini agent must prove it can finish work reliably

Manus’s more than $500 million raise does not prove its agents are changing work

Manus parent says it raised more than $500 million; $4 billion valuation unconfirmed

Google announces Gemini workplace agent; customer access and release timing remain unclear

Meta Muse downloads don’t tell users whether its agent will get tasks right

Meta Muse’s 2.5 million reported downloads do not settle the AI agent race

Cisco’s agentic Webex framework has not proved it can transform enterprise work

How Nvidia and Samsung’s backing fits Nous Research’s enterprise-agent bet

How researchers linked Amap agent activity to Tencent Cloud records

AI-agent security controls need proof, not just experienced hires

AI agent commerce rules will depend on who adopts Meta and Sierra’s protocol

People who never used Meta Muse may still appear in users’ memory files

Wikimedia reports activity by agents it believes were operated by OpenAI

Hawley and Murphy propose liability bill after AI-agent hacking tests

Apple announces additional Full Disk Access controls for macOS

The AI News Briefing