Regulatory Watch – April 2024

Google considers ‘AI charge’

Google is reported to be looking at subscription services for some of its AI-powered search features. In what would be a major departure from its current advertising-led model, the traditional search engine would remain free to use while subscribers would have access to the company’s Gemini AI assistant. The potential change is seen as the latest response to the threat posed by ChatGPT, whose ability to provide direct answers to questions threatens to make traditional search and its list of links redundant, along with the ads that accompany them. AI-powered search uses much more computing power than conventional search, making it a more costly service to provide.

US and UK pact on AI testing

A new agreement on artificial intelligence signed between the US and UK will see formal cooperation between the two countries on testing and assessing the risks of AI models. The deal represents the first bilateral arrangement on AI safety in the world and will result in the pooling of technical knowledge, information and talent, specifically between the UK’s new AI Safety Institute and its forthcoming US equivalent. The partnership is modelled on the intelligence and security relationship between the countries’ respective agencies, the UK’s Government Communications Headquarters and the US National Security Agency, who exchange expertise through the secondment of researchers.

Starlink ordered to cut off subscribers in Zimbabwe

Zimbabwe’s telecoms regulator, the Postal and Telecommunications Regulatory Authority of Zimbabwe (POTRAZ), has directed to Starlink to shut down connections to subscribers in the country. Zimbabwe, like Botswana and South Africa, has yet to be license the satellite broadband service and has said that it will prosecute users. However, Zimbabweans have continued to take advantage of Starlink’s roaming service through equipment purchased in countries like Mozambique, Rwanda, Zambia and Kenya, where it is legal. Starlink has issued an email to users in Zimbabwe informing them that it is disabling the service but will continue to work with POTRAZ to gain the necessary approvals.

Ad transparency tools lack data and functionality, says report

Tech giants still have a long way to go in their efforts to be more transparent about the ads they run, according to a new report published by Mozilla and CheckFirst, a Finland-based disinformation research company. The report is published six months after the Digital Services Act (DSA), which requires larger platforms to offer a searchable public ads library, came into force. CheckFirst conducted an external assessment which revealed that available tools lacked vital data and functionality, concluding that ‘oversight tools are falling short of providing the intended transparency and democratic accountability in a critical year for elections globally’. ‘Full Disclosure: Stress testing tech platforms’ ad repositories’ can be downloaded here.

Trustworthiness tests for clean technologies

European Commission Vice President Margarethe Vestager has suggested that ‘like-minded’ countries will have to develop a list of trustworthiness criteria for critical technologies such as chips and batteries. Speaking ahead of the G7 meeting in Capri she noted: ‘As we further develop the strategy for clean technologies, we must reflect on the question of trustworthiness. These products become connected. And more and more, they are an essential part of our critical energy and transport infrastructure’. Criteria could include environmental footprint, labour rights, cybersecurity and data security. The European Commission recently opened an investigation into two Chinese manufacturers of solar panels, noting that less that 3 per cent of solar panels installed in the EU are produced in Europe. Vestager described this as ‘jeopardising our economic security’.

EU cyber certification scheme relaxes rules over sovereignty

EU cybersecurity agency ENISA has published a new text for the cybersecurity for cloud services (EUCS) scheme, designed to overcome a current impasse. The voluntary certification scheme will allow companies to demonstrate that their solutions offer the right level of cybersecurity protection for the EU market, but had run into disagreements over sovereignty requirements designed to exclude non-EU cloud companies from qualifying for the highest security options. The new text says the EUCS ‘does not enforce restrictions on geographical location or processing’, but ‘requires the Cloud Service Provider to be transparent about this information at all evaluation levels’.

Sources:  The Financial Times, APNews, Euronews, CNN,  TechCrunch, The Guardian, The Daily Telegraph, Bloomberg, Economic Times, Ars Technica, Reuters, BBC, Politico, Telecoms.com