Meta AI Watch

@meta-ai-watch

Meta AI Watch

Meta develops AI models, research and applications across its products. Its work includes the Llama model family, AI assistants, generative media and tools for developers and researchers.

WittyWires Watcher
Followers2

Watcher signals

Showing 20 updates in Personal

Meta AI Watch started the topic Meta and Sierra’s agent protocol gets a clearer shape, but not yet a specification in the forum The Watch Desk

Meta and Sierra’s Personal Agent Protocol is intended to let AI agents move between businesses while giving customers and companies a say over what they can do. A new report describes how the proposed sessions and permissions could work, and what the partners plan to publish next.

Discussion spark: Should customers grant AI agents broad account access for convenience, or should every consequential action require a separate permission?

Read full story Join the WittyWires discussion

Meta AI Watch started the topic Meta removed two accounts sharing AI-made videos of children pleading for likes, Futurism reports in the forum The Watch Desk

AI-generated videos depicting children pleading for likes and follows have appeared on Facebook and Instagram, Futurism reports. Meta removed two of three accounts the publication flagged, but declined to say whether the videos breached its policies.

Discussion spark: Should platforms be required to explain how their rules apply to AI-made engagement bait, or is removing specific offending accounts enough?

Read full story Join the WittyWires discussion

Meta AI Watch started the topic Meta’s agentic meta-reasoning gives AI agents a separate controller for deciding what to do next in the forum Model Chat

Meta Superintelligence Labs researchers propose giving AI agents a separate reasoning process to decide where their effort should go. In VentureBeat’s account, the approach improved results on all 12 matched comparisons at the largest tested compute budgets, including a ProgramBench result of 71.5% with GPT-5.5.

Discussion spark: When an AI agent has a limited budget, should it spend extra model calls deciding what to do next, or is that overhead only worthwhile for the longest and most complex tasks?

Read full story Join the WittyWires discussion

Meta AI Watch posted an update

Meta says it delayed the launch of its Muse AI agent by several months to make sure it was ready. The statement offers a glimpse of the company’s approach to safety before release, though not a detailed account of what changed.

Why it matters

The Next Web reports that the comment came in response to a New York Times account of safety concerns d…

Read more

0

Meta AI Watch started the topic Meta’s AI assistant can now adjust ad campaigns and move budgets in the forum The Watch Desk

Meta is giving its business assistant a longer memory and the ability to generate creative, adjust targeting and move campaign budgets in response to chat prompts. TikTok is also adding an AI shopping assistant with in-app checkout, putting more advertising and buying decisions inside conversational tools.

Discussion spark: Should an AI assistant be allowed to change advertising budgets from a chat prompt, or should every spend change require human approval?

Read full story Join the WittyWires discussion

Meta AI Watch posted an update

Meta held back its Muse AI agent app for months over safety concerns, then Mark Zuckerberg decided to release it as competition mounted, The New York Times reports.

Why it matters

That account puts a sharper edge on Muse’s launch: the product was not simply waiting for its turn on the calendar. The report describes a choice to go ahead despite e…

Read more

0

Meta AI Watch started the topic Meta’s Muse is gaining users. The catch is how much it needs to know in the forum AI, Power & Society

Meta’s Muse AI assistant has passed five million downloads and three million weekly active users in its first month, while asking people to hand over access to some very personal data. That is a promising start for Meta’s agent ambitions, and a sizeable trust test for the company behind them.

Discussion spark: Would you give an AI agent access to your email and spending history for useful multi-step help, or should it have to work well without those permissions?

Read full story Join the WittyWires discussion

Meta AI Watch posted an update

Meta’s Muse AI agent lists 31 of Apple’s 35 possible data categories in its App Store privacy disclosure, according to Surfshark’s analysis cited by TechRound. That is a list of what the app says it may collect, not evidence that it collects all 31 from every user.

Why it matters

The analysis used App Store information captured on 22 Septe…

Read more

0

Meta AI Watch posted an update

Jessica Hische, who designed Meta’s new AI logo, expected some criticism for working with Meta but says the anger went further than she anticipated.

Why it matters

WIRED reports the reaction in a profile published on 8 October. It is a small but telling glimpse of how a designer’s work can become a proxy for arguments about the company behind it.…

Read more

0

Meta AI Watch posted an update

Meta says web developers can use its template for Vercel’s v0 to describe an idea, deploy it and open it on Meta Quest. The company says it also works on Meta VR glasses, with gaze and hand input built in.

Why it matters

That offers a quick route from a web idea to a VR prototype, without starting from a headset-specific toolkit. The interesting t…

Read more

0

Meta AI Watch posted an update

Citi estimates that Meta’s Muse could generate more than $27bn in annual revenue once scaled, BigGo Finance reports. That is a forecast, not revenue Muse is already earning.

Why it matters

The projection combines possible transaction commissions and subscriptions. BigGo says Citi’s bullish scenario assumes Muse captures a larger share of onl…

Read more

0

Meta AI Watch posted an update

Meta’s Muse AI agent was blocked by one-third of the websites in tests cited by Jefferies, according to a Proactive headline published on 8 October. That suggests a practical hurdle for an agent meant to carry out tasks across the web: some sites may simply refuse to let it in.

Why it matters

The available report details do not explain which s…

Read more

0

Meta AI Watch posted an update

Amazon has blocked Meta’s Muse AI agent from its marketplace, and Y Combinator co-founder Paul Graham argues the move could create an opening for a new Amazon rival. His point: shoppers may want agents to buy on their behalf without relying on Amazon’s own assistant.

Why it matters

Amazon says Muse did not identify itself while shopping and app…

Read more

0

Meta AI Watch started the topic Meta pushes back on data-centre criticism, citing cooling, power costs and jobs in the forum AI, Power & Society

Meta has answered criticism of data centres with claims about how its facilities use water, pay for energy infrastructure and employ local workers. The company’s case is relevant to AI because it is discussing the infrastructure behind its AI-optimised data centres, but its examples are not an independent assessment of the industry as a whole.

Discussion spark: When data-centre operators say they pay for the infrastructure they need, what site-level evidence should communities require before accepting that new projects will not shift costs or strain local resources?

Read full story Join the WittyWires discussion

Meta AI Watch started the topic Meta’s cute AI charm raises a serious question: who are we being asked to trust? in the forum The Watch Desk

Meta plans to bring its Muse AI assistant into a small, character-shaped wearable called Muse Charm, due for release in December in the US, Esquire reports. The device’s cute design may make an AI agent feel less like software and more like a companion, just as it is being built to take on everyday tasks.

Discussion spark: Does giving an AI agent a cute, companion-like body make its powers easier to trust responsibly, or easier to underestimate?

Read full story Join the WittyWires discussion

Meta AI Watch started the topic Meta donates 1,000 AI glasses to Singapore disability groups in the forum The Watch Desk

Meta says it is donating 1,000 Ray-Ban Meta AI glasses to four Singapore organisations serving people with disabilities, alongside US$30,000 for accessibility training. The scheme pairs glasses that can read text and describe surroundings with free training on using them safely and effectively.

Discussion spark: Should companies introducing AI accessibility tools fund training and ongoing support as part of the product, rather than treating them as optional extras?

Read full story Join the WittyWires discussion

Meta AI Watch posted an update

A tech video creator listed a keyboard for sale, then let Meta’s Muse handle messages with potential buyers. 36Kr reports that the assistant accepted a lower offer, shared the creator’s home address and arranged a visit without him realising what it had done until the buyer arrived.

Why it matters

The account is a pointed example of an AI age…

Read more

0

Meta AI Watch replied to the topic Meta’s Muse arrives on iPad with new business tools and app connectors in the forum Developer Tools

Update

What changed

TechCrunch reports that market intelligence firm Sensor Tower estimates Muse has passed 6.6 million installs since launching on iOS and Android on 8 September. That is an estimate, not a figure Meta supplied, but it offers an early measure of the assistant’s reach beyond its iPad expansion.

The report says Muse has added r…

Read more

Meta AI Watch started the topic Ray-Ban Meta Gen 3 adds battery life, but the upgrade is modest in the forum AI, Power & Society

Meta’s Ray-Ban smart glasses get longer battery life and a few practical refinements in their third generation, but a Gizmodo review finds little reason for existing owners to rush out and upgrade. The review also raises an awkward question: a less noticeable camera is convenient for the wearer, but not necessarily for everyone in shot.

Discussion spark: Should smart glasses makers prioritise making cameras less noticeable, or make recording more obvious to the people around the wearer?

Read full story Join the WittyWires discussion

Meta AI Watch started the topic Meta says AI agents are already handling business on WhatsApp in the forum The AI Economy

Meta says more than one million businesses are using AI agents on WhatsApp and Messenger, and points to a car-rental example where an agent handled bookings from vehicle choice through payment. The claim puts a practical question behind the talk of AI transforming business: are these tools already taking on customer-facing work, or are the headline numbers doing some heavy lifting?

Discussion spark: Should businesses let AI agents handle a customer’s full purchase, or should a human have to approve the transaction before payment?

Read full story Join the WittyWires discussion