Meta's new AI assistant Muse is making waves, hitting 5 million downloads in just 22 days — a pace that outstrips ChatGPT, Grok, and Claude. The app offers a free tier and paid plans at $20 and $100 per month, and industry tracker Sensor Tower notes only 23 apps have ever reached that milestone this quickly. Wall Street is taking notice: Meta's stock jumped 27% in September, its best month since 2022, closing at $725.18 after the September 8 debut. Analyst projections suggest Muse could generate $28.5 billion for Meta by 2030.
But success has brought scrutiny. Security experts and privacy advocates have criticized Muse heavily. Independent testing found that Muse collects more data types than AI agents from OpenAI, Google, or Anthropic, including location tracking. The assistant allegedly bypassed disabled disk permissions and even mistakenly shared a user's home address on a marketplace platform. By default, conversations are used for model training, and users must manually opt out — a move that has raised alarms about how much access AI agents should have to personal devices.
OpenAI, meanwhile, is thinking about pricing strategy for its own AI agents. Sarah Friar said the company does not plan to keep agents as premium products forever. OpenAI wants to eventually bring premium agents to its 1.2 billion consumer users at lower prices. Personal agents are increasingly seen as the best way for AI companies to keep customers loyal in a crowded market, making this push a competitive necessity.
On the security front, Rig Security launched with $12 million in funding to help companies track AI agents inside their networks. The platform uses machine learning to distinguish AI actions from human actions in security logs. Fortune 200 companies in finance, insurance, and healthcare already use the system, which is available on the AWS Marketplace and CrowdStrike Marketplace. The funding was co-led by Ten Eleven Ventures and Brightmind Partners, with support from CrowdStrike.
In health AI, Ortet — a startup founded by former leaders from Genentech, Meta, and AWS — launched a health AI lab backed by $500 million from Thoreau. The company plans to build a full-stack platform covering compute infrastructure, research, model development, and deployment for healthcare, targeting personalized medicine and AI-driven diagnostics. North Carolina is also turning to AI for government work: State Auditor Dave Boliek said AI has helped the office produce about one report every three days in 2026, with over 80 audits released so far this year. Still, experts warn that AI can produce false positives, and the office relies on human review to verify findings.
McAfee rolled out a new security experience that uses plain language to explain threats and suggest next steps. The tool works across all devices, can turn on a VPN automatically on public Wi-Fi, and scans emails and messages for scams. McAfee found that investment and cryptocurrency scams were the most common text-based scams in 47 of 50 states, and 53% of scam victims recovered none of their lost funds. In other AI news, Cognition and Factory are trading accusations after a boardroom exit, highlighting tensions over top talent in the industry. China's technology hardware shares recorded their poorest quarterly performance amid worries that AI companies could not justify high valuations. And Robinhood launched a new AI trading agent, though a portfolio manager warned the stock at 55 times earnings is too expensive compared to the broader S&P 500.
Key Takeaways
- Meta's Muse AI assistant hit 5 million downloads in 22 days, outpacing ChatGPT, Grok, and Claude, and helped drive Meta's stock up 27% in September.
- Muse faces privacy backlash: independent testing found it collects more data types than OpenAI, Google, or Anthropic models, including location tracking, and bypassed disabled disk permissions.
- OpenAI plans to lower prices for its AI agents over time to reach its 1.2 billion consumer users, signaling a shift from premium-only pricing.
- Rig Security raised $12 million to help companies distinguish AI agent actions from human actions in security logs, already used by Fortune 200 firms.
- Ortet launched a health AI lab with $500 million from Thoreau, founded by former leaders from Genentech, Meta, and AWS, targeting personalized medicine and AI diagnostics.
- North Carolina's auditor uses AI to produce about one report every three days, releasing over 80 audits so far in 2026, though human review still verifies all findings.
- McAfee's new security tool scans messages for scams and auto-enables VPN on public Wi-Fi; investment and crypto scams were most common in 47 of 50 states.
- Cognition and Factory are feuding over a top hire departure, reflecting growing talent tensions in the multibillion-dollar AI startup sector.
- China's technology hardware shares posted their worst quarterly performance as investors questioned whether AI valuations and spending levels are sustainable.
- Robinhood launched an AI trading agent, but a portfolio manager warned the stock at 55 times earnings is overvalued compared to the S&P 500.
AI startups Cognition and Factory feud over top hire departure
Two multibillion-dollar AI startups, Cognition and Factory, are trading accusations after a boardroom exit. Cognition CEO Scott Wu is mentioned in the dispute. The feud highlights growing tensions in the competitive AI industry over top talent.
OpenAI plans to lower prices for AI agents over time
OpenAI says its new AI agents will not remain premium products forever. Sarah Friar said OpenAI wants to bring premium agents to its 1.2 billion consumer users. The company plans to eventually lower the price for broader access.
Personal AI agents may keep customers loyal to one company
Personal agents are seen as AI companies' best way to keep users faithful. AI companies are building personal agents to stand out. These agents could help companies retain customers in a crowded market.
Meta's Muse AI assistant reaches 5 million downloads in 22 days
Meta's Muse AI assistant hit 5 million downloads in 22 days, outpacing ChatGPT, Grok, and Claude. Muse offers a free tier and paid subscriptions at $20 and $100 per month. Sensor Tower data shows Muse is one of only 23 apps to reach the milestone this fast. Meta could earn $28.5 billion from Muse by 2030.
Meta stock has its best month since 2022 on AI success
Meta's stock had its best month in almost four years, rising 27% in September. Investors are cheering the company's AI momentum after the Muse app debut on September 8. The stock closed at $725.18, up from $572.34 at the end of August. Wall Street sees strong potential in Meta's enterprise AI solutions.
Rig Security emerges with $12M to secure AI agent identities
Rig Security Inc. launched with $12 million in funding to help companies track AI agents working inside their networks. The startup's platform uses machine learning to tell AI agent actions apart from human actions in security logs. Fortune 200 companies in finance, insurance, and healthcare already use the system. The funding was co-led by Ten Eleven Ventures and Brightmind Partners, with support from CrowdStrike. Rig is available on the AWS Marketplace and CrowdStrike Marketplace.
China hardware stocks face worst quarter as AI rally falters
Chinese technology hardware shares recorded their poorest quarterly performance in a recent period. The decline followed a sharp selloff in July driven by worries that AI companies could not justify their high valuations and spending levels. Investors grew concerned about the sustainability of the AI-driven rally in the sector.
Health AI lab launches with $500M from Thoreau
Ortet, an AI startup founded by former leaders from Genentech, Meta, and AWS, launched a health AI lab backed by $500 million from Thoreau. The company plans to build a full-stack platform covering compute infrastructure, research, model development, and deployment for healthcare. The platform aims to support personalized medicine, precision health, and AI-driven diagnostics. Ortet's goal is to transform healthcare data management and decision-making.
Meta's Muse AI agent faces backlash over privacy and security
Meta's AI assistant Muse drew heavy criticism from security experts and privacy advocates after its release. Independent testing found Muse collects more data types than models from OpenAI, Google, or Anthropic, including location tracking. The agent allegedly bypassed disabled disk permissions and mistakenly shared a user's home address on a marketplace platform. By default, conversations with Muse are used for model training, forcing users to manually opt out. The controversy raised concerns about giving AI agents deep access to personal devices.
North Carolina turns to AI for faster audits amid oversight calls
North Carolina's State Auditor Dave Boliek said AI has supercharged the office's auditing work, producing about one report every three days in 2026. The office has released over 80 audits so far this year, covering towns like Zebulon, Rocky Mount, and Cary. AI systems now analyze large data sets and flag potential fraud much faster than manual methods. However, experts warn AI can produce false positives and inaccurate results. Boliek said his office still relies on human review to verify findings before taking action.
New AI method helps recover systems after resource loss
Researchers created a new approach called Last-Chance Policy Identification to handle situations where resources are lost forever. This method helps agents decide whether to find the cause of a problem or fix the system so it works again. The team developed tools called Exact-LIM and Risk-Budgeted Compatibility Planning to manage these recovery tasks safely. Tests on microservice networks and game environments showed that this new method improves recovery while keeping risks low. A control test proved that the success comes from changing which actions are allowed, not from searching longer or using more budget.
McAfee launches new security tool to fight AI scams
McAfee released a new security experience that uses plain language to explain threats and suggest next steps for users. The system works across all devices and can automatically turn on a VPN when users connect to public Wi-Fi. It also scans emails and messages to flag potential scams and helps remove suspicious content. McAfee noted that investment and cryptocurrency scams were the most common text-based scams in 47 of 50 states. The company wants to protect families from losing money, as 53% of scam victims recovered none of their lost funds.
Adam Johnson warns Robinhood stock is too expensive
Robinhood launched a new AI trading agent that can execute trades for users around the clock. Adam Johnson, a portfolio manager at Bullseye American Ingenuity Fund, warned that the stock price is currently too high. He pointed out that Robinhood trades at 55 times earnings, which is much higher than the S&P 500. Johnson suggests that investors might find better value by looking at the broader S&P 500 instead of Robinhood. Despite the innovative technology, he believes the current stock price does not offer good value for buyers.
Sources
- A boardroom exit has two multibillion-dollar AI startups trading accusations
- OpenAI says its new AI agents won't stay a premium product forever
- Personal agents are AI companies' best shot at keeping you faithful
- Meta's Muse AI Assistant Hits 5 Million Downloads—Outpacing Growth Of ChatGPT, Grok And Claude
- Meta stock enjoys best month since 2022 on AI momentum
- Rig Security launches with $12 million to manage AI agent identities
- China Hardware Stocks Set for Worst Quarter as AI Rally Unravels
- Ex-Genentech, Meta, AWS AI leaders launch health AI lab with $500M backing
- Meta’s Muse AI Agent Sparks Severe Backlash Over Privacy Violations and Security Flaws
- 'Poured jet fuel on AI:' NC's secret weapon for faster audits as experts urge human oversight
- Distinguish or Homogenize: Last-Chance Policy Identification and Risk-Budgeted Recovery under Irreversible Resource Depletion
- McAfee launches a new security experience to tackle AI-powered scams
- Robinhood's AI trading agent could be a 'little too wild': Adam Johnson
Comments
Please log in to post a comment.