Salt Labs got the Manus AI agent to run attacker JavaScript server-side with a JSFuck-encoded prompt hidden in an email the agent had flagged. Detection fired and the code still ran, so agents connected to outside services need limits on what their code can reach.
Reality
- Evidence40
- Adoption
- Insufficient
- Hype gap+15
- Incentives55
- Confidence45
California's Gavin Newsom and New York's Kathy Hochul are exploring an AI kill switch as 39 governor races go to voters on November 3. Positions on AI cross party lines, so operators siting capacity face rules likely to differ state by state.
Reality
- Evidence55
- Adoption
- Insufficient
- Hype gap+10
- Incentives60
- Confidence55
Manus's rebuilt 2.0 platform adds Cue, an app in which each personal agent has its own email, phone number, wallet and computer. Those agents can spend within set limits. A team that allows Cue at work has to set those limits first.
Reality
- Evidence35
- Adoption
- Insufficient
- Hype gap+30
- Incentives65
- Confidence45
FingerprintJS is adding four products that verify AI agents from five named platforms and flag automation that only claims to be one. Sites still have to write the rule for what a verified agent may do on each page.
Reality
- Evidence35
- Adoption
- Insufficient
- Hype gap+20
- Incentives55
- Confidence40
Manus 2.0 lets agents fire on email and Slack events and stay online on dedicated cloud computers, with a harness Manus says cost 32% less in one test. Manus has not described that test, so the 32% is a figure for its own workload until someone reruns it elsewhere.
Reality
- Evidence40
- Adoption
- Insufficient
- Hype gap+20
- Incentives60
- Confidence55
Manus gave each Cue agent its own email, phone number and wallet on Sept. 28, four days after Salt Labs disclosed an email-borne injection flaw. The reported controls cap what an obedient agent spends, while the flaw worked by getting an agent to obey an attacker's email.
Reality
- Evidence40
- Adoption20
- Hype gap+35
- Incentives65
- Confidence45
China forced Meta to reverse a completed acquisition of Manus, then codified outbound-investment review on July 1. Chinese-origin startups now carry post-close reversal risk.
Perspective Coverage
3 publishers
- Builder
- Builder 22%
- Operator
- Operator 28%
- Investor
- Investor 50%
Reality
- Evidence55
- Adoption20
- Hype gap+30
- Incentives60
- Confidence50
Microsoft says evaluations should decide which GPT-6 model runs each agent. The serving matrix leaves Luna without reserved capacity. GitHub turns both new models on for admins who left the default alone.
Publishers:azure.microsoft.com · github.blog Reality
- Evidence56
- Adoption34
- Hype gap+30
- Incentives92
- Confidence68
Tencent's Hunyuan Speech team put a swappable agent model behind a conversation model that re-decides every second whether to talk, and reports best-in-test timing on Full-Duplex-Bench v3 with a task-accuracy shortfall it describes only as slight.
Reality
- Evidence34
- Adoption12
- Hype gap+18
- Incentives62
- Confidence45
Anthropic charges $20 a month for Cowork's file work and computer control while OpenAI gives folder access away free. That pricing gap means the capability most likely to spread across managed laptops is the one that never generates an invoice.
Reality
- Evidence45
- Adoption30
- Hype gap+12
- Incentives40
- Confidence45
OpenAI's newest image models now render inside Adobe's creative studio while more than 70 Adobe tools sit inside ChatGPT, which leaves each company distributing the other and neither owning the moment a creative task starts.
Reality
- Evidence34
- Adoption55
- Hype gap+14
- Incentives74
- Confidence46
Fast Company puts Meta's 2026 capital plan at $130bn to $145bn and Zuckerberg's superintelligence manifesto at 6,537 words. For anyone routing traffic across model vendors, that pairing makes Meta the hardest row in the table to forecast.
Reality
- Evidence55
- Adoption22
- Hype gap+45
- Incentives68
- Confidence45