Tech Brief - The general purpose AI codes of practice row, Meta accused of violating DMA

“If the General Purpose AI Codes of Practice drafting process is not multi-stakeholder, including civil society, academics, and independent experts, this could mean an industry-led process; essentially Big Tech writing their own rules.”

– A civil society stakeholder who declined to be named on the Codes of Practice process

Story of the week: The European Commission plans to allow the providers of ChatGPT-artificial intelligence (AI) models, to write codes of practice that will determine their compliance in the short to medium term, with civil society in a consultation role. (Read more)

Don’t miss: The European Commission accused Meta of violating its digital competition rules, the Digital Markets Act (DMA), in its “pay or OK” model, officials told Euractiv on Monday. The model gives users a choice between consenting to using their personal data for targeted advertising or paying for an ad-free experience on Meta platforms. “The DMA isn’t about whether there is advertising or not on a service. It is about the amount and the type of data which is available for this advertising. That means there are other forms of advertising, which do not rely on your full personal information being shared,” said an official. (Read more)

Also this week:

Before we start: If you just can’t get enough tech analysis, tune in on our weekly podcast.

Compliance assessments in risk classification. AI-based cybersecurity and emergency services components in internet-connected devices are expected to be classified as high-risk in the context of the AI Act, according to a European Commission document seen by Euractiv. The document on the interplay between the 2014 Radio Equipment Directive (RED) and the AI Act is the first known interpretation of how the Act will treat AI-based safety components, laying down the logic that could be used to classify other types of products as high-risk.

Milestone AI bill advanced in California. As federal legislators can’t agree on AI legislation, California is taking matters into their own hands. A bipartisan bill passed a committee vote on Tuesday, and is headed for more committee votes before eventually a vote on the floor of the state’s Senate. It requires AI companies to safety test AI models before deployment and implement safety measures if model training costs more than $100 million (€ 92.4 million), or compute use surpasses a threshold of 10^26 operations, ten times the compute requirement for posing a systemic risk according to the AI Act. The bill advances in face of considerable pushback from venture capital and tech companies saying it will place an impossible burden on developers.

Brazil halts Meta’s AI data mining. On Tuesday, Brazil banned Meta’s practice of using public posts and images to train AI models, unless users go through a tedious opt-out process. The Data Protection Authority made this decision in the wake of strong pushback in Europe, and a Human Rights Watch report raising concerns over the ability to create malicious deepfakes of Brazilian children.

Google’s guzzling carbon. The tech giant noted an increase of 48% in greenhouse gas emissions in 2023 compared to 2019, according to Google’s sustainability report, released on Wednesday. The company said it is because of increased AI computing being plugged into its data centres. At the same time, Google said 64% of its global energy consumption was carbon free in 2023, same as 2022.

Vestager on Apple AI. Apple’s decision not to launch its own AI features in the EU is a “stunning declaration” of its anticompetitive behaviour, European Commission Vice-President Margrethe Vestager said last Thursday.

Defence industry on EUCS. The aerospace, security and defence industry is set to voice concerns about an EU-wide cloud cybersecurity certification scheme (EUCS). The Association of Space and Defence (ASD), which represents 4,000 companies, according to its website, in a position paper already sent to the Commission and to be published soon, will be calling for provisions on data localisation and contractual guarantees, to ensure industrial data does not fall into foreign hands, Giorgio Mosca, chair of the ASD cybersecurity task force, told Euractiv.

China’s push for equitable AI. The UN adopted a China-led resolution pushing for a “free, open, inclusive and non-discriminatory” business environment between developing and wealthy nations to create AI, on Monday. A few days later, China’s Premier Li Qiang called for narrowing the “intelligence gap” in AI and stronger international cooperation in the technology’s development, in a speech in Shanghai, SCMP reported.

Apple’s fine appeal. Details on Apple’s appeal of a 1.8 billion fine under the DMA were released in the Official Journal of the EU on Monday. The company is contesting its definition as a dominant player as well as the finding that its anti-steering provisions are abusive. The Commission fined Apple in March over what it said are abusive practices against streaming providers.

E-commerce investigations. The European Commission formally requested information, on Friday 28 June, from online marketplaces Temu and Shein on measures taken to comply with online safety regulation Digital Services Act (DSA).

X fine is set to face a formal warning from the EU Commission for not properly addressing illegal content on its platform, Bloomberg reported on Thursday. The Commission opened an investigation into Elon Musk’s X under the DSA in December.

Disinformation in French elections. Russian disinformation campaigns are targeting social media platforms to destabilise the French political scene – currently in the middle of a legislative campaign – a study by the French National Centre for Scientific Research (CNRS) focused exclusively on X reads.

Florida and Texas don’t get to regulate social media. The US Supreme Court voted to hold off Florida and Texas efforts to regulate user-generated content on social media platforms, the AP reported on Monday. With varying details, the cases aimed at addressing conservative concerns that social media leans left. The highest court of the US sent the cases back to state judiciaries, largely on free speech grounds.

[Edited by Rajnish Singh]

Thanks for reading. Be sure to spread the word and come and say hello on X.