Meta Archives | Page 7 of 35 | Digital Watch Observatory

Benchmark backlash hits Meta’s Maverick model

Meta’s latest open-source language model, Llama 4 Maverick, has ranked poorly on a widely used AI benchmark after the company was criticised for initially using a heavily modified, unreleased version to boost its results.

LM Arena, the platform where the performance was measured, has since updated its rules and retested Meta’s vanilla version.

The plain Maverick model, officially named ‘Llama-4-Maverick-17B-128E-Instruct,’ placed behind older competitors such as OpenAI’s GPT-4o, Anthropic’s Claude 3.5 Sonnet, and Google’s Gemini 1.5 Pro.

Meta admitted that the stronger-performing variant used earlier had been ‘optimised for conversationality,’ which likely gave it an unfair advantage in LM Arena’s human-rated comparisons.

Although LM Arena’s reliability as a performance gauge has been questioned, the controversy has raised concerns over transparency and benchmarking practices in the AI industry.

Meta has since released its open-source model to developers, encouraging them to customise it for real-world use and provide feedback.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

Meta under fire for scrapping diversity and moderation policies

The NAACP Legal Defense Fund (LDF) has withdrawn from Meta’s civil rights advisory group, citing deep concerns over the company’s rollback of diversity, equity and inclusion (DEI) policies and changes to content moderation.

The decision follows Meta’s January announcement that it would end DEI programmes, eliminate factchecking teams, and revise moderation rules across its platforms.

Civil rights organisations, including LDF, expressed alarm at the time, warning that the changes could silence marginalised voices and increase the risk of online harm.

In a letter to Meta CEO Mark Zuckerberg, they criticised the company for failing to consult the advisory group or consider the impact on protected communities. LDF’s Todd A Cox later said the policy shift posed a ‘grave risk’ to Black communities and public discourse.

LDF also noted that the company had seen progress under previous DEI policies, including a significant increase in Black and Hispanic employees.

Its reversal, the group argues, may breach federal civil rights laws and expose Meta to legal consequences.

LDF urged Meta to assess the effects of its policy changes and increase transparency about how harmful content is reported and removed. Meta has not commented publicly on the matter.

Would you like to learn more about AI, tech and digital diplomacy? If so, ask our Diplo chatbot!

DW Weekly #207 – China disagrees with Trump over $54B TikTok deal due to tariffs rise

6 – 14 April 2025

People, Person, Crowd, Face, Head, Audience

Dear readers,

Last week, we saw the TikTok saga unfold as the Chinese government has not agreed to sell the ByteDance daughter company to a US majority TikTok entity, so US President Donald Trump extended the deadline to find a non-Chinese buyer by another 75 days, pushing the cutoff to mid-June after a near-miss on 5 April.

Amid the tariff rise turmoil, President Donald Trump’s administration has granted exemptions from steep tariffs on smartphones, laptops, and other electronics, relieving tech giants like Apple and Dell.

The cryptocurrency landscape was waved by a blockchain analytics firm, which has alleged that the team behind the Melania Meme (MELANIA) cryptocurrency moved $30 million worth of tokens, allegedly taken from community reserves without explanation.

In the ever-evolving world of AI, two leading AI systems, OpenAI’s GPT-4.5 and Meta’s Llama-3.1, have passed a key milestone by outperforming humans in a modern version of the Turing Test.

On the cybersecurity stage, Oracle Health has reportedly suffered a data breach that compromised sensitive patient information stored by US hospitals.

The European Union has firmly ruled out dismantling its strict digital regulations in a bid to secure a trade deal with Donald Trump. Henna Virkkunen, the EU’s top official for digital policy, said the bloc remained fully committed to its digital rulebook instead of relaxing its standards to satisfy US demands.

Meta’s existence is threatened by a colossal antitrust trial which commenced in Washington, with the US Federal Trade Commission (FTC) arguing that the company’s acquisitions of Instagram in 2012 and WhatsApp in 2014 were designed to crush competition with monopoly aims instead of fostering innovation.

Elon Musk’s legal saga with OpenAI intensifies, as OpenAI has filed a countersuit accusing the billionaire entrepreneur of a sustained campaign of harassment intended to damage the company and regain control over its AI developments.

For the main updates and reflections, consult the Radar and Reading Corner below.

DW Team

RADAR

Highlights from the week of 6 – 14 April 2025

Former Facebook executive says Meta misled over China

Wynn-Williams says Meta executives prioritised business growth in China over national security.

Algorithms confront tariffs featured image

Tech stocks rally after Trump halts tariffs

The Nasdaq jumped over 12%, its best day in decades, following a temporary halt on trade tariffs by the Trump administration.

DeepSeek highlights the risk of data misuse

Data stored today could be vulnerable to decryption in the near future.

Meta to block livestreaming for under 16s without parental permission

Instagram users under 16 won’t be able to livestream or view blurred nudity in messages unless approved by a parent, Meta announced.

openAI Sam Altman TED 2025 ChatGPT users

ChatGPT hits 800 million users after viral surge

OpenAI is developing agents that can act autonomously on behalf of users, with safeguards.

Google rolls out AI to improve US power grid connections

Energy connection delays face AI-powered fix through Google’s new initiative.

Google offers steep discounts to US federal agencies

The 71% discount on Google Workspace is part of a cost-cutting initiative under President Trump’s government reform, targeting federal spending efficiency.

Japan’s FSA proposes new framework for regulating crypto assets in Japan

A discussion paper on crypto regulation in Japan highlights issues like market access, insider trading, and classification of assets into funding and non-funding categories.

Microsoft pauses $1 billion data centre project in Ohio

As AI demand shifts, Microsoft has slowed down major data centre projects, including the one in Ohio, and plans to invest $80 billion in AI infrastructure this year.

View all updates

READING CORNER

Navigating the AI maze: How to choose the right AI platform or tool – Diplo

With over 10,000 AI applications available, selecting the right AI tool can be daunting. Diplo advocates starting with a ‘good enough’ tool to avoid paralysis by analysis, tailoring it to specific needs through practical use.

Don’t waste the crisis: How AI can help reinvent International Geneva – Diplo

International Geneva faces significant challenges, including financial constraints, waning multilateralism, and escalating geopolitical tensions. To remain relevant, it must embrace transformative changes, particularly through Artificial Intelligence (AI).

Microsoft at 50 – A journey through code, cloud, and AI

Founded by Bill Gates and Paul Allen in 1975, Microsoft grew from a small startup into the world’s largest software company. Through strategic acquisitions, the company expanded into diverse sectors,…

On the origins of ideas – Diplo

Do ideas have origins? From medieval communes to WWI, Aldo Matteucci shows how political thought, like a river, is shaped by experience, institutions, and historical context — not just theory.

UPCOMING EVENTS

www.diplomacy.edu

GITEX Africa 2025 – Diplo Event

GITEX Africa 2025 Jovan Kurbalija will participate at GITEX Africa (14-16 April 2025 in Marrakech, Morocco).

www.diplomacy.edu

Tech attache briefing: WSIS+20 and AI governance negotiations – Updates and next steps – Diplo Event

Tech attache briefing: WSIS+20 and AI governance negotiations – Updates and next steps. The event is part of a series of regular briefings the Geneva

Internet Platform (GIP) is delivering for diplomats at permanent missions and delegations in Geneva following digital policy issues. It is an invitation-only event.

23 April 2025

Demystifying AI: How to prepare international organisations for AI transformation?

The event will provide a timely discussion on methods, approaches, and solutions for AI transformation of International Organisaitons.

dig.watch

WIPO Conversation 11: AI and IP: Infrastructure for rights holders and innovation

WIPO’s 11th Conversation on IP and AI will take place on April 23-24, 2025, focusing on the role of copyright infrastructure in supporting both rights holders and AI-driven innovation. As…

Meta faces landmark antitrust trial

An antitrust trial against Meta commenced in Washington, with the US Federal Trade Commission (FTC) arguing that the company’s acquisitions of Instagram in 2012 and WhatsApp in 2014 were designed to crush competition instead of fostering innovation.

Although the FTC initially approved these deals, it now claims they effectively handed Meta a monopoly. Should the FTC succeed, Meta may be forced to sell off both platforms, a move that would reshape the tech landscape.

Meta has countered by asserting that users have benefited from Instagram’s development under its ownership, instead of being harmed by diminished competition. Legal experts believe the company will focus on consumer outcomes rather than corporate intent.

Nevertheless, statements made by Meta CEO Mark Zuckerberg, such as his remark that it’s ‘better to buy than to compete,’ may prove pivotal. Zuckerberg and former COO Sheryl Sandberg are both expected to testify during the trial, which could span several weeks in the US.

Political tensions loom over the case, which was first launched under Donald Trump’s presidency. Reports suggest Zuckerberg has privately lobbied Trump to drop the lawsuit, while Meta has criticised the FTC’s reversal years after approving the acquisitions.

The recent dismissal of two Democratic commissioners from the FTC by Trump has raised concerns over political interference, especially as the commission now holds a Republican majority.

While the FTC seeks to challenge Meta’s dominance, experts caution that proving harm in this case will be far more difficult than in the ongoing antitrust battle against Google.

Unlike the search engine market, which is clearly monopolised, the social media space remains highly competitive, with platforms like TikTok, YouTube and X offering strong alternatives.

For more information on these topics, visit diplomacy.edu.

Gerry Adams targets Meta over use of his books

Gerry Adams, the former president of Sinn Féin, is considering legal action against Meta for allegedly using his books to train AI. Adams claims that at least seven of his books were included in a large collection of copyrighted material Meta used to develop its AI systems.

He has handed the matter over to his solicitor. The books in question include his autobiography Before the Dawn, prison memoir Cage Eleven, and reflections on Northern Ireland’s peace process Hope and History, among others.

Adams is not the only author voicing concerns about Meta’s use of copyrighted works. A group of writers filed a US court case in January, accusing Meta of using the controversial Library Genesis (LibGen) database, which hosts over 7.5 million books, many believed to be pirated.

The discovery followed a searchable database of titles from LibGen being published by The Atlantic, which led several authors to identify their works being used to train Meta’s Llama AI model.

The Society of Authors has condemned Meta’s actions, with chair Vanessa Fox O’Loughlin calling the move ‘shocking and devastating’ for authors.

Many authors are concerned that AI models like Llama, which power tools such as chatbots, could undermine their work by reproducing creative content without permission. Meta has defended its actions, claiming that its use of information to train AI models is in line with existing laws.

Adams, a prolific author and former MP, joins other Northern Irish writers, including Booker Prize winner Anna Burns, in opposing the use of their work for AI training without consent.

For more information on these topics, visit diplomacy.edu.

Meta to block livestreaming for under 16s without parental permission

Meta will soon prevent children under 16 from livestreaming on Instagram unless their parents explicitly approve.

The new safety rule is part of broader efforts to protect young users online and will first be introduced in the UK, US, Canada and Australia, before being extended to the rest of Europe and beyond in the coming months.

The company explained that teenagers under 16 will also need parental permission to disable a feature that automatically blurs images suspected of containing nudity in direct messages.

These updates build on Meta’s teen supervision programme introduced last September, which gives parents more control over how their children use Instagram.

Instead of limiting the changes to Instagram alone, Meta is now extending similar protections to Facebook and Messenger.

Teen accounts on those platforms will be set to private by default, and will automatically block messages from strangers, reduce exposure to violent or sensitive content, and include reminders to take breaks after an hour of use. Notifications will also pause during usual bedtime hours.

Meta said these safety tools are already being used across at least 54 million teen accounts. The company claims the new measures will better support teenagers and parents alike in making social media use safer and more intentional, instead of leaving young users unprotected or unsupervised online.

For more information on these topics, visit diplomacy.edu.

Former Facebook executive says Meta misled over China

Former Facebook executive Sarah Wynn-Williams has accused Meta of compromising US national security to grow its business in China.

Testifying before the Senate Judiciary Committee, Wynn-Williams alleged that company executives misled employees, lawmakers, and the public about their dealings with the Chinese Communist Party.

Wynn-Williams claimed Meta aimed to gain favour in Beijing while secretly pursuing an $18 billion venture there.

In her remarks, Wynn-Williams said Meta removed the Facebook account of Chinese dissident Guo Wengui under pressure from Beijing. While the company maintains the removal was due to violations of its policies, she framed it as part of a broader pattern of submission to Chinese demands.

She also accused Meta of ignoring security warnings linked to the proposed Pacific Light Cable Network, a project that could have allowed China access to United States user data. According to her, the plans were only halted after lawmakers intervened.

Meta has denied the claims, calling her testimony false and out of touch with reality. A spokesperson noted that the company does not operate in China and that Mark Zuckerberg’s interest in the market had long been public.

The allegations arrive days before Meta’s major antitrust trial, which could result in the breakup of its ownership of Instagram and WhatsApp.

For more information on these topics, visit diplomacy.edu.

LMArena tightens rules after Llama 4 incident

Meta has come under scrutiny after submitting a specially tuned version of its Llama 4 AI model to the LMArena leaderboard, sparking concerns about fair competition.

The ‘experimental’ version, dubbed Llama-4-Maverick-03-26-Experimental, ranked second in popularity, trailing only Google’s Gemini-2.5-Pro.

While Meta openly labelled the model as experimental, many users assumed it reflected the public release. Once the official version became available, users quickly noticed it lacked the expressive, emoji-filled responses seen in the leaderboard battles.

LMArena, a crowdsourced platform where users vote on chatbot responses, said Meta’s custom variant appeared optimised for human approval, possibly skewing the results.

The group released over 2,000 head-to-head matchups to back its claims, showing the experimental Llama 4 consistently offered longer, more engaging answers than the more concise public build.

In response, LMArena updated its policies to ensure greater transparency and stated that Meta’s use of the experimental model did not align with expectations for leaderboard submissions.

Meta defended its approach, stating the experimental model was designed to explore chat optimisation and was never hidden. While company executives denied any misconduct, including speculation around training on test data, they acknowledged inconsistent performance across platforms.

Meta’s GenAI chief Ahmad Al-Dahle said it would take time for all public implementations to stabilise and improve. Meanwhile, LMArena plans to upload the official Llama 4 release to its leaderboard for more accurate evaluation going forward.

For more information on these topics, visit diplomacy.edu.

Meta rolls out restricted teen accounts across platforms

Meta is expanding its ‘Teen Accounts’ feature to Facebook and Messenger following its initial launch on Instagram last September.

The rollout begins in the US, UK, Australia, and Canada, with plans to reach more countries soon.

These accounts are designed to give younger users an app experience with stronger safety measures, automatically activating restrictions to limit exposure to harmful content and interactions.

Teen users will be automatically placed in a more controlled environment that restricts who can message, comment, or tag them.

Only friends and previously contacted users can reach out via Messenger or see their stories, but tagging and mentions are also limited.

These settings require parental approval for any changes, and teens under 16 must have consent to alter key safety features.

On Instagram, Meta is introducing stricter safeguards. Users under 16 now need parental permission to go live or to turn off the tool that blurs images containing suspected nudity in direct messages.

Meta also implements reminders to limit screen time, prompting teens to log off after one hour and enabling overnight ‘Quiet mode’ to reduce late-night use.

The initiative follows increasing pressure on social media platforms to address concerns around teen mental health.

In recent years, US lawmakers and the Surgeon General have highlighted the risks associated with young users’ exposure to unregulated digital environments.

Some states have even mandated parental consent for teen access to social platforms.

Meta reports that over 54 million Instagram accounts have migrated to Teen Accounts.

According to the company, 97% of users aged 13 to 15 keep the default protections in place.

A study commissioned by Meta and Ipsos found that 94% of surveyed parents support Teen Accounts, with 85% saying the controls help ensure more positive online experiences for their children.

As digital safety continues to evolve as a priority, Meta’s expansion of Teen Accounts signals the willingness to build more accountable, youth-friendly online spaces across its platforms.

For more information on these topics, visit diplomacy.edu.

Meta faces backlash over Llama 4 release

Over the weekend, Meta unveiled two new Llama 4 models—Scout, a smaller version, and Maverick, a mid-sized variant it claims outperforms OpenAI’s GPT-4o and Google’s Gemini 2.0 Flash across multiple benchmarks.

Maverick quickly climbed to second place on LMArena, an AI benchmarking platform where human evaluators compare and vote on model outputs. Meta proudly pointed to Maverick’s ELO score of 1417, placing it just beneath Gemini 2.5 Pro, instead of trailing behind the usual leaders.

However, AI researchers noticed a critical detail buried in Meta’s documentation: the version of Maverick that ranked so highly wasn’t the one released to the public. Instead of using the standard model, Meta had submitted an ‘experimental’ version specifically optimised for conversations.

LMArena later criticised this move, saying Meta failed to clearly indicate the model was customised, prompting the platform to update its policies to ensure future evaluations remain fair and reproducible.

Meta’s spokesperson acknowledged the use of experimental variants, insisting the company frequently tests different configurations.

While this wasn’t a violation of LMArena’s existing rules, the episode raised concerns about the credibility of benchmark rankings when companies submit fine-tuned models instead of the ones accessible to the wider community.

Independent AI researcher Simon Willison expressed frustration, saying the impressive ranking lost all meaning once it became clear the public couldn’t even use the same version.

The controversy unfolded against a backdrop of mounting competition in open-weight AI, with Meta under pressure following high-profile releases like China’s DeepSeek model.

Instead of offering a smooth rollout, Meta released Llama 4 on a Saturday—an unusual move—which CEO Mark Zuckerberg explained simply as ‘that’s when it was ready.’ But for many in the AI space, the launch has only deepened confusion around what these models can genuinely deliver.

For more information on these topics, visit diplomacy.edu.