Search “Meta”

Meta updates business terms to allow using user content to train AI

Meta updated its business terms this month to allow it to use content from associated domains, web pages or URLs to develop, improve, research, train and maintain Meta products, sparking discontent among artists and small business owners who rely on Instagram and Facebook. The effective date of the terms was postponed from October 30 to November 20, and Meta subsequently removed the word "train" from the list of permitted uses, limiting the purposes to "advertising and business products." Users in the UK and EU can currently submit an objection form in the Instagram Privacy Center, and some users say they have received emails from Meta confirming that their objections were approved; these options are not displayed in the United States and other places.

Meta deploys Catching JiTTests to the production environment

Meta has published the production deployment results of Catching JiTTests: the system uses large language models to automatically generate executable tests for code changes awaiting merge, exposing defects before the code is merged. The system employs two workflows, intent-aware and dodgy diff, and uses an LLM-as-Judge ensemble together with the rule-based method RubFake to filter false positives, while reducing manual item-by-item review through a human-on-the-loop approach. Meta says 589 engineers have participated, with a true positive rate of 60%; in the first half of 2026, it caught 74 severe privacy SEVs at the diff stage, reducing the overall incidence of such issues by 20%.

Meta deploys Catching JiTTests to production

Meta has published the production deployment results of Catching JiTTests: the system uses large language models to automatically generate executable tests for code changes awaiting merge, exposing defects before the code is merged. The system employs two workflows, intent-aware and dodgy diff, and uses an LLM-as-Judge ensemble together with the rule-based method RubFake to filter false positives, while reducing manual item-by-item review through a human-on-the-loop approach. Meta says 589 engineers have participated, with a true positive rate of 60%; in the first half of 2026, it caught 74 severe privacy SEVs at the diff stage, reducing the overall incidence of such issues by 20%.

View story:Meta deploys Catching JiTTests to the production environment

Meta updates its commercial terms to allow using user content to train AI

Meta updated its business terms this month to allow it to use content from associated domains, web pages or URLs to develop, improve, research, train and maintain Meta products, sparking discontent among artists and small business owners who rely on Instagram and Facebook. The effective date of the terms was postponed from October 30 to November 20, and Meta subsequently removed the word "train" from the list of permitted uses, limiting the purposes to "advertising and business products." Users in the UK and EU can currently submit an objection form in the Instagram Privacy Center, and some users say they have received emails from Meta confirming that their objections were approved; these options are not displayed in the United States and other places.

View story:Meta updates business terms to allow using user content to train AI

Meta introduces PrivacyLib privacy control library

On October 28, 2026, Meta's engineering team released PrivacyLib, a general-purpose privacy library that runs inside the clients of hundreds of data and processing systems and performs billions of checks per second. The library only needs to be integrated once, and new privacy requirements can reuse that integration as modules, eliminating the need to reimplement them in each system. In addition, the newly added Agent Observability module can observe data access events of internal AI agents, and systems that have already integrated the library can enable it without additional modifications.

Meta introduces PrivacyLib privacy control library

On October 28, 2026, Meta's engineering team released PrivacyLib, a general-purpose privacy library that runs inside the clients of hundreds of data and processing systems and performs billions of checks per second. The library only needs to be integrated once, and new privacy requirements can reuse that integration as modules, eliminating the need to reimplement them in each system. In addition, the newly added Agent Observability module can observe data access events of internal AI agents, and systems that have already integrated the library can enable it without additional modifications.

View story:Meta introduces PrivacyLib privacy control library

Meta plans to raise up to $95 billion through an equity offering

On October 26, 2026, Meta Platforms announced plans to raise up to $95 billion through an equity offering, to be used for AI infrastructure, computing power, and personal superintelligence. The first round of transactions totaled $35 billion, of which about two-thirds came from common stock, and $12.5 billion in Series A depositary shares were issued; subsequently, it plans to raise up to an additional $60 billion through an at-the-market offering. NVIDIA, AMD, and Broadcom expressed interest in subscribing to a combined total of up to $5 billion in the Class A offering, but this is non-binding. On October 27, AMD announced a deep strategic partnership with Meta and invested $1 billion in Meta's equity offering. On October 28, Meta's $25 billion common stock offering was about five times oversubscribed.

AMD announces deep strategic partnership with Meta

On October 26, 2026, Meta Platforms announced plans to raise up to $95 billion through an equity offering, to be used for AI infrastructure, computing power, and personal superintelligence. The first round of transactions totaled $35 billion, of which about two-thirds came from common stock, and $12.5 billion in Series A depositary shares were issued; subsequently, it plans to raise up to an additional $60 billion through an at-the-market offering. NVIDIA, AMD, and Broadcom expressed interest in subscribing to a combined total of up to $5 billion in the Class A offering, but this is non-binding. On October 27, AMD announced a deep strategic partnership with Meta and invested $1 billion in Meta's equity offering. On October 28, Meta's $25 billion common stock offering was about five times oversubscribed.

View story:Meta plans to raise up to $95 billion through an equity offering

Meta Shares AI Networking Innovations at OCP Summit

At the 2026 OCP Global Summit, Meta unveiled networking innovations supporting its latest AI models and its gigawatt-scale Prometheus cluster. The first liquid-cooled network rack introduced by Meta, IcePack, debuted alongside the 102.4T switch IceCube (128×800G, available with either Broadcom Tomahawk 6C or NVIDIA Spectrum-6). For scale-out, the Flattened Planar Fabric connects each GPU to eight network planes, simplifying the Clos architecture from 3 tiers to 2; for scale-across, Meta introduced BAG, a petabit-class super backbone connecting buildings, campuses and regions. Meta also introduced Tahan, an MTIA 300 Ethernet scale-up switch based on Broadcom Tomahawk5, and open-sourced its MetaRoCE software implementation, with the first user-space version, libsoftmetaroce, now available on GitHub.

Meta shares AI networking innovations at OCP Summit

At the 2026 OCP Global Summit, Meta unveiled networking innovations supporting its latest AI models and its gigawatt-scale Prometheus cluster. The first liquid-cooled network rack introduced by Meta, IcePack, debuted alongside the 102.4T switch IceCube (128×800G, available with either Broadcom Tomahawk 6C or NVIDIA Spectrum-6). For scale-out, the Flattened Planar Fabric connects each GPU to eight network planes, simplifying the Clos architecture from 3 tiers to 2; for scale-across, Meta introduced BAG, a petabit-class super backbone connecting buildings, campuses and regions. Meta also introduced Tahan, an MTIA 300 Ethernet scale-up switch based on Broadcom Tomahawk5, and open-sourced its MetaRoCE software implementation, with the first user-space version, libsoftmetaroce, now available on GitHub.

View story:Meta Shares AI Networking Innovations at OCP Summit

Meta acquires AI data analytics platform Sundial

Meta has acquired the AI data analytics platform Sundial. The platform was founded by Julie Zhuo, former Vice President of Product Design at Meta, and Chandra Narayanan, who previously led data analytics at Instagram and Sequoia Capital. Its core capability is enabling enterprises to ask questions of complex data in natural language through a chatbot plug-in and quickly obtain analytical conclusions. According to Phoenix Technology, after the integration, Narayanan will be responsible for Meta's data analytics tools and AI agent-related work. IT Home also reported that in July this year, Meta appointed Alex Schultz as its first Chief Data Officer and formed the DATA department. No further developments have been disclosed at this time.

Meta acquires AI data analytics platform Sundial

Meta has acquired the AI data analytics platform Sundial. The platform was founded by Julie Zhuo, former Vice President of Product Design at Meta, and Chandra Narayanan, who previously led data analytics at Instagram and Sequoia Capital. Its core capability is enabling enterprises to ask questions of complex data in natural language through a chatbot plug-in and quickly obtain analytical conclusions. According to Phoenix Technology, after the integration, Narayanan will be responsible for Meta's data analytics tools and AI agent-related work. IT Home also reported that in July this year, Meta appointed Alex Schultz as its first Chief Data Officer and formed the DATA department. No further developments have been disclosed at this time.

View story:Meta acquires AI data analytics platform Sundial

Meta's Muse AI agent launches in Japan

Meta's personal AI agent Muse first landed in the Japanese market on October 23, 2026, after the product had previously launched in North America. According to Wallstreetcn, Muse launched in North America on September 8 and reached 10 million downloads in six weeks; Testing Catalog, citing a post by Meta Chief AI Officer Alexandr Wang, said Muse had previously launched in the United States and Canada and is now entering Japan. The same report also mentioned that Wang, in an interview with Nikkei, said Muse is free for ordinary users and does not run ads, while advanced features and enterprise services will be charged, and the company also plans to take a commission from shopping transactions facilitated by the agent. There is currently no further operational data after the Japan market launch.

Meta paper proposes RSD method to let agents learn from unlabeled trajectories

Meta published a paper proposing the RSD method, which lets AI agents, after knowing the task outcome, review each step and learn from historical attempts that no one labeled as right or wrong. Experimental results reported in the paper show: on ScienceWorld, Qwen3-14B's success rate rose from 12.9% to 20.6%; on ALFWorld, Qwen3-8B reached 76.6%, surpassing all tested methods, including those that received rewards or standard answers. The authors also recommend retaining agents' unlabeled run records (including failures) and regenerating the review notes each round, because when notes were frozen at round 3, ALFWorld success rate stalled at 59.7%, whereas with notes updated each round it was 67.9%. Currently this work is released as a paper, with no subsequent progress yet.

Meta paper proposes RSD method to enable agents to learn from unlabeled historical trajectories

Meta published a paper proposing the RSD method, which lets AI agents, after knowing the task outcome, review each step and learn from historical attempts that no one labeled as right or wrong. Experimental results reported in the paper show: on ScienceWorld, Qwen3-14B's success rate rose from 12.9% to 20.6%; on ALFWorld, Qwen3-8B reached 76.6%, surpassing all tested methods, including those that received rewards or standard answers. The authors also recommend retaining agents' unlabeled run records (including failures) and regenerating the review notes each round, because when notes were frozen at round 3, ALFWorld success rate stalled at 59.7%, whereas with notes updated each round it was 67.9%. Currently this work is released as a paper, with no subsequent progress yet.

View story:Meta paper proposes RSD method to let agents learn from unlabeled trajectories

Meta expects 2026 capital expenditures of $135 billion to $145 billion

In its third-quarter earnings report released on October 21, Meta forecast 2026 capital expenditures of $135 billion to $145 billion, to continue investing in AI infrastructure and new product development. The company said its AI investments have already generated incremental advertising revenue through stronger user engagement, better recommendations and more efficient monetization, and it expects these investments and new product launches to support revenue growth in 2027. The earnings report showed third-quarter revenue of $64.6 billion, up 26% year over year, including advertising revenue of $63.06 billion; total costs and expenses for the quarter were $51.61 billion, up 68% year over year, including $9.89 billion in charges related to a settlement with 52 U.S. state attorneys general over litigation concerning teen mental health and social media.

Meta stock plunged 6.98% to $648.85

Meta stock closed down 6.98% at $648.85 per share on October 22, falling back to early September levels and giving up all the gains brought by the Muse agent. A report Meta submitted to the U.S. Securities and Exchange Commission the same day showed that, as of September 30, 2026, its total irrevocable contractual commitments stood at $454.52 billion, mainly involving third-party cloud computing capacity, servers and network infrastructure, data center energy services, and Reality Labs consumer hardware, of which approximately $45.85 billion and $121.41 billion mature in 2026 and 2027, respectively. There are currently no subsequent developments.

Meta expects 2026 capital expenditures of $135 billion to $145 billion

In its third-quarter earnings report released on October 21, Meta forecast 2026 capital expenditures of $135 billion to $145 billion, to continue investing in AI infrastructure and new product development. The company said its AI investments have already generated incremental advertising revenue through stronger user engagement, better recommendations and more efficient monetization, and it expects these investments and new product launches to support revenue growth in 2027. The earnings report showed third-quarter revenue of $64.6 billion, up 26% year over year, including advertising revenue of $63.06 billion; total costs and expenses for the quarter were $51.61 billion, up 68% year over year, including $9.89 billion in charges related to a settlement with 52 U.S. state attorneys general over litigation concerning teen mental health and social media.

View story:Meta expects 2026 capital expenditures of $135 billion to $145 billion

Meta releases third-quarter financial report

On October 22, 2026, Meta released its third-quarter financial report. According to Zhi Dongxi reporting, the company's revenue was 64.601 billion USD, up 26% year over year, and net profit was 9.860 billion USD, up 264% year over year, but after including 9.89 billion USD in costs related to adolescent mental health litigation, operating profit fell 37% year over year; as of September 30, global employees totaled 66,775, down 15% year over year, about 12,000 fewer. On the earnings call, Meta set its personal AI agent Muse as the future core; Zuckerberg said Muse was downloaded and set up more than 10 million times in the United States and Canada within one month of launch, with a quarter of users using it daily; because computing costs have prevented a global launch, it plans to improve computing cost-effectiveness by 10 times next year, and plans expenditures of 135 billion to 145 billion USD in 2026. There is currently no further subsequent development confirmed by factual evidence.

Meta stock plunges 6.98% to $648.85

Meta stock closed down 6.98% at $648.85 per share on October 22, falling back to early September levels and giving up all the gains brought by the Muse agent. A report Meta submitted to the U.S. Securities and Exchange Commission the same day showed that, as of September 30, 2026, its total irrevocable contractual commitments stood at $454.52 billion, mainly involving third-party cloud computing capacity, servers and network infrastructure, data center energy services, and Reality Labs consumer hardware, of which approximately $45.85 billion and $121.41 billion mature in 2026 and 2027, respectively. There are currently no subsequent developments.

View story:Meta stock plunged 6.98% to $648.85