Meta fixes Muse bug on Mac after 500,000 downloads

Key Takeaways

1Muse, Meta's AI agent, surpassed 500,000 downloads on MacOS in one week

2A zero-day bug detected by Patrick Wardle has been fixed, but there is still a disagreement on the exploitation conditions

3The MacOS update includes the patch, and no further action is required

Why it matters : The rapid adoption of Muse comes with security and trust challenges, as Meta Connect must present the company's vision on AI amid heightened privacy concerns.

Meta is gearing up for Meta Connect as Muse, its AI agent on Mac, surpasses 500,000 installations in just one week. A fix has been released following the discovery of a zero-day vulnerability, amid a public disagreement over its exploitation conditions. The update on MacOS includes the patch, with no further action required.

Meta Connect arrives against a backdrop of a troubled past and public doubts

Meta Connect is scheduled for Wednesday, with an anticipated address from Mark Zuckerberg on AI strategy and associated hardware. Historically, the event has focused on Meta's smart glasses, the long-term viability of which is being questioned due to growing public discontent, with some labeling them as "pervert glasses" because of their recording capabilities. Meta also carries a history marked by the Cambridge Analytica scandal in 2016, which revealed the inappropriate use of data from millions of Facebook users, as well as recent proceedings concerning online child protection. Furthermore, public trust in AI companies has weakened, as AI researchers express concerns about the extreme risks associated with this technology.

Zero-day bug detected, fix released, and controversy over exploitation

A serious zero-day bug was patched this week in Muse. Patrick Wardle, founder of Objective-See, was the first to detect the vulnerability and described it on X, showing that an attacker could redirect dictations, audio recordings, and Muse access tokens to a malicious server instead of Meta's. David Singleton, head of Meta's Superintelligence Labs, confirmed on Tuesday that the flaw had been fixed. He clarified that exploiting the bug required the prior presence of malware on the user's computer, a claim contested by Wardle, who asserts that remote exploitation is possible.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

MacOS update recommended, patch included, and no additional steps required

Muse is available as a MacOS application, and an update allows users to utilize the latest version that includes the security fix. No additional action is required from the user. David Singleton, representing Meta, states that the company aims for high transparency regarding privacy and security in Muse to maintain trust.

Rapid adoption and competitive ecosystem, constrained by sensitive permissions

According to internal data, the MacOS application of Muse exceeded 500,000 downloads in its first week, in a market already populated by agents from OpenAI, SpaceXAI, and Anthropic. For tasks like booking a dinner, Muse needs to access the computer, connect to the Internet, and potentially use credit card information, which fuels privacy and security concerns associated with generative AI. No tech company has fully addressed all potential dangers of AI, whether it be hacking or loss of control over personal data. Muse, an agentic assistant embodied by a cartoon character on Instagram, fits into this landscape, with Meta having a longer history than most startups in the sector. Patrick Wardle also has a recognized track record in discovering bugs in Meta's software.