EU tech regulators are closely monitoring the messaging app Telegram as it approaches a significant usage milestone that could subject it to stricter regulations under the EU’s Digital Services Act (DSA). The DSA, which came into effect last year, imposes tougher obligations on major tech companies to control illegal and harmful content on their platforms.
Telegram reported having 41 million EU users in the six months leading up to February, just shy of the 45 million threshold that would categorise it as a very large online platform (VLOP). If it reaches this threshold, Telegram will need to comply with more stringent regulations. The European Commission has confirmed ongoing communication with Telegram to monitor its user growth and compliance.
Currently, 18 online platforms, including giants like Google, Amazon, Apple, Meta, and TikTok, are already classified as VLOPs under the DSA. These platforms are required to adhere to rigorous standards to ensure safer online environments, reflecting the EU’s commitment to mitigating online risks and safeguarding digital spaces.
According to Ireland’s Data Protection Commission, leading global internet companies are working closely with the EU regulators to ensure their AI products comply with the bloc’s stringent data protection laws. This body, which oversees compliance for major firms like Google, Meta, Microsoft, TikTok, and OpenAI, has yet to exercise its full regulatory power over AI but may enforce significant changes to business models to uphold data privacy.
AI introduces several potential privacy issues, such as whether companies can use public data to train AI models and the legal basis for using personal data. AI operators must also guarantee individuals’ rights, including the right to have their data erased and address the risk of AI models generating incorrect personal information. Significant engagement has been noted from tech giants seeking guidance on their AI innovations, particularly large language models.
Following consultations with the Irish regulator, Google has already agreed to delay and modify its Gemini AI chatbot. While Ireland leads regulation due to many tech firms’ EU headquarters being located there, other EU regulators can influence decisions through the European Data Protection Board. AI operators must comply with the new EU AI Act and the General Data Protection Regulation, which imposes fines of up to 4% of a company’s global turnover for non-compliance.
Why does it matter?
Ireland’s broad regulatory authority means that companies failing to perform due diligence on new products could be forced to alter their designs. As the EU’s AI regulatory landscape evolves, these tech firms must navigate both the AI Act and existing data protection laws to avoid substantial penalties.
The Biden administration is pressing major technology companies to intensify their efforts to reduce antisemitic content on their platforms. Representatives from Alphabet, Meta, Microsoft, TikTok, and X met with US special envoy Deborah Lipstadt to discuss strategies for monitoring and combating antisemitism. Lipstadt emphasised the need for each company to assign a policy team member to address the issue, conduct specialised training to identify antisemitism, and publicly report trends in anti-Jewish content.
TikTok supported the meeting, highlighting their ongoing efforts and commitment to learning from experts. However, Alphabet, Microsoft, Meta, and X have yet to respond to requests for comment on the matter. The US administration is also calling for enhanced training to help platform staff recognise subtle antisemitic messages and distinguish between legitimate criticism of the Israeli government and hate speech directed at Jews.
The push from the administration comes amid a global increase in antisemitism following the 7 October attack by Hamas on southern Israel and the subsequent Israeli military response in Gaza. While the tech companies have not yet committed to voluntary measures, Lipstadt remains hopeful that they will take action soon to address this pressing issue.
The European Centre for Digital Rights, or Noyb, has filed a complaint against OpenAI, claiming that ChatGPT fails to provide accurate information about individuals. According to Noyb, the General Data Protection Regulation (GDPR) mandates that information about individuals be accurate and that they have full access to this information, including its sources. However, OpenAI admits it cannot correct inaccurate information on ChatGPT, citing that factual accuracy in large language models remains an active research area.
Noyb highlights the potential dangers of ChatGPT’s inaccuracies, noting that while such errors may be tolerable for general uses like student homework, they are unacceptable when they involve personal information. The organisation cites a case where ChatGPT provided an incorrect date of birth for a public figure, and OpenAI refused to correct or delete the inaccurate data. Noyb argues this refusal breaches the GDPR, which grants individuals the right to rectify incorrect data.
Furthermore, Noyb points out that the EU law requires all personal data to be accurate, and ChatGPT’s tendency to produce false information, known as ‘hallucinations’, constitutes another violation of the GDPR. Data protection lawyer Maartje de Graaf emphasises that the inability to ensure factual accuracy can have serious consequences for individuals, making it clear that current chatbot technologies like ChatGPT are not compliant with the EU laws regarding data processing.
Noyb has requested that the Austrian data protection authority (DSB) investigate OpenAI’s data processing practices and enforce measures to ensure compliance with the GDPR. The organisation also seeks a fine against OpenAI to promote future adherence to data protection regulations.
The European Commission announced on Tuesday that Meta Platforms has introduced measures to combat disinformation ahead of the EU elections. Meta has launched 27 real-time visual dashboards, one for each EU member state, to enable third-party monitoring of civic discourse and election activities.
This development comes after the European Commission investigated Meta last month for allegedly breaching EU online content regulations. The investigation highlighted concerns over Meta’s Facebook and Instagram platforms failing to address disinformation and deceptive advertising adequately.
While the formal procedures against Meta continue, the European Commission stated that it would closely monitor the implementation of these new features to ensure their effectiveness in curbing disinformation.
OpenAI has established a Safety and Security Committee to oversee the training of its next AI model, the company announced on Tuesday. CEO Sam Altman will lead the committee alongside directors Bret Taylor, Adam D’Angelo, and Nicole Seligman. The committee makes safety and security recommendations to OpenAI’s board.
The committee’s initial task is to review and enhance OpenAI’s existing safety practices over the next 90 days, after which it will present its findings to the board. Following the board’s review, OpenAI plans to share the adopted recommendations publicly. This move follows the disbanding of OpenAI’s Superalignment team earlier this month, which led to the departure of key figures like former Chief Scientist Ilya Sutskever and Jan Leike.
Other members of the new committee include technical and policy experts Aleksander Madry, Lilian Weng, and head of alignment sciences John Schulman. Newly appointed Chief Scientist Jakub Pachocki and head of security Matt Knight will also be part of the committee, contributing to the safety and security oversight of OpenAI’s projects and operations.
Fake accounts discussing the US presidential election are increasing on the social media platform X, according to a report by Cyabra, an Israeli tech company specialising in AI-driven analysis. The report found that 15% of accounts praising former President Donald Trump and criticising President Joe Biden are fake, while 7% of accounts praising Biden and criticising Trump are fake.
Cyabra’s study analysed posts on X over two months, starting 1 March, focusing on popular hashtags and sentiment. The analysis showed a tenfold increase in fake accounts during March and April. Specifically, 12,391 out of 94,363 pro-Trump accounts and 803 out of 10,065 pro-Biden accounts were found to be bogus.
The report also noted that fake pro-Trump accounts appear to be part of a coordinated campaign pushing messages like ‘Vote for Trump’ and ‘Biden is the worst president the US has ever had,’ while fake pro-Biden accounts did not show coordinated activity.
Although X did not respond to requests for comments, Elon Musk recently announced efforts to purge bots and trolls from the platform, including testing a ‘Not a Bot’ program in New Zealand and the Philippines.
Why does it matter?
Since Russia’s interference in the 2016 election, social media platforms have faced increased scrutiny. With the upcoming election on 5 November, Cyabra’s findings on fake accounts are a cause of alarm for election officials and misinformation experts. The situation is even more concerning, given X’s history of downplaying the presence of fake accounts on its platform. According to Reuters, in May 2022, Twitter claimed that fewer than 5% of its daily active users were ‘false or spam’ based on an internal review. However, Cyabra estimated that 13.7% of Twitter profiles were inauthentic.
A Russian rights group, OVD-Info, reported that YouTube has threatened to block one of its channels in Russia, called Kak Teper, which discusses Ukraine war and political issues and has 100,000 subscribers.
Reuters reports that YouTube’s warning followed a complaint from Russian regulator Roskomnadzor, claiming the content violated information technology laws. OVD-Info is negotiating with YouTube and Google, labelling the potential block as political censorship.
YouTube did not specify which law was violated and did not respond to inquiries about this case but confirmed the reinstatement of videos from other opposition channels. Blocking YouTube entirely in Russia could be unpopular due to its large user base of tens of millions of monthly users.
Why does it matter?
OVD-Info’s Kak Teper might become the first entire human rights channel banned on YouTube, warns Natalia Krapiva from Access Now, noting the growing threat to civil society’s presence on the platform. While Russia has blocked most foreign social media, YouTube has managed to avoid a ban, but not without consequences, as it has been consistently fined for hosting content deemed illegal by Russian authorities.
The EU’s privacy watchdog task force has raised concerns over OpenAI’s ChatGPT chatbot, stating that the measures taken to ensure transparency are insufficient to comply with data accuracy principles. In a report released on Friday, the task force emphasised that while efforts to prevent misinterpretation of ChatGPT’s output are beneficial, they still need to address concerns regarding data accuracy fully.
The task force was established by Europe’s national privacy watchdogs following concerns raised by authorities in Italy regarding ChatGPT’s usage. Despite ongoing investigations by national regulators, a comprehensive overview of the results has yet to be provided. The findings presented in the report represent a common understanding among national authorities.
Data accuracy is a fundamental principle of the data protection regulations in the EU. The report highlights the probabilistic nature of ChatGPT’s system, which can lead to biassed or false outputs. Furthermore, the report warns that users may perceive ChatGPT’s outputs as factually accurate, regardless of their actual accuracy, posing potential risks, especially concerning information about individuals.
The Biden administration is urging tech and financial industries to combat the proliferation of AI-generated sexual abuse images, the Time reports. Generative AI tools have made it easy to create explicit deepfakes, often targeting women, children, and LGBTQ+ individuals, with little recourse for the victims. The White House calls for voluntary cooperation from companies to implement measures to stop these nonconsensual images, while no federal legislation addresses the issue.
Biden’s chief science adviser, Arati Prabhakar, noted the rapid increase in such abusive content and the need for companies to take responsibility. A document shared with the Associated Press outlines actions for various stakeholders, including AI developers, financial institutions, cloud providers, and app store gatekeepers, to restrict the monetisation and distribution of explicit images, particularly those involving minors. The administration also stressed the importance of stronger enforcement of terms of service and better mechanisms for victims to remove nonconsensual images online.
Why does it matter?
Last summer, major tech companies committed to AI safeguards, followed by an executive order from Biden to ensure AI development prioritises public safety, including measures to detect AI-generated child abuse imagery. However, high-profile incidents, like AI-generated deepfakes of Taylor Swift and the rise of such images in schools, reveal and urgent need for action and the potential insufficiency of voluntary commitments from companies. Recently, Forbes reported that AI-generated images of young girls in provocative outfits are spreading on TikTok and Instagram, drawing inappropriate comments from older men and raising concerns about potential exploitation.
