News Briefing 20-24 May

Autonomy in weapons systems, and AI in security and defence

• International: Use of AI in weaponry and war demands great care over ‘accountability gap’: This week, the International Affairs Department of the Bishops’ convened at a conference where the Catholic church reiterated its ‘opposition to the development and use of weapons systems’ that may cause harm. The Church released a document that sets out thirteen practical action points that include ‘a call to place emerging technology such as AI at the service of humanity, ensuring that we learn the lessons of history.’ Pope Francis also stated that ‘the workings of AI and its potential are beyond the ability of most of us to understand and appreciate, [it] has proven both exciting and disorienting.’ 
• International: Autonomous weapons: Canada urged to act: Interviewed by La Presse last week,  Mary Wareham, Deputy Director of the Crisis, Conflict and Arms Division of Human Rights Watch shares some thoughts and expresses that the government of Canada needs to urgently respond to the critical issue of autonomous weapon systems globally. She states that in the past Canada played a pivotal role and was ‘leading actor in the development’ of the historic anti personnel  landmine treaty.’ 
• International: Before applying AI to war, we’d better think about its limitations: Niusha Shafiabady, an associate professor at the Faculty of Science and Technology at Charles Darwin University provides commentary for the Australian Strategic Policy Institute. Shafiabady says ‘ There is no doubt that AI will become increasingly prominent in war. But policymakers need to be aware of its shortcomings as well as its capabilities.’

AI, algorithms and autonomy in the wider world

• International: AI Seoul Summit: World leaders agree to launch network of safety institutes: On 21 May, South Korea and the UK jointly hosted a virtual AI Safety Summit where global Leaders also signed up to the wider Seoul Declaration. ‘During this conference, a number of states declared the importance of ‘enhanced international cooperation to develop human-centric, trustworthy AI.’
• International: Meta AI chief says large language models will not reach human intelligence: The Financial Times covers an interview with Yann LeCun, the chief AI scientist at Meta where he strongly articulated ‘against relying on advancing LLMs in the quest to make human-level intelligence, as these models can only answer prompts accurately if they have been fed the right training data and are, therefore, intrinsically unsafe.’
• UK: UK watchdog looking into Microsoft AI taking screenshots: The UK Information Commissioner’s Office (ICO) has raised concerns over recall, a new AI feature by Microsoft which collects screenshots from the user’s computer and stores it locally in memory. While Microsoft maintains that they don’t have access to these screenshots as they are saved locally, experts have indicated that this ‘could be a privacy nightmare.’ 
• US: A Widely Used Criminal Justice Algorithm For Assessing Child Pornography Recidivism Is Flawed: This piece by Nicholas Scurich, chair of the Department of Psychological Science at the University of California and Daniel Krauss, professor in the department of psychological science at Claremont McKenna College critically analyzes the Child Pornography Offender Risk Tool (CPORT) used in the State of Georgia. The authors criticize the algorithmic tool owing to minimal and non diverse training data used, as well as its ‘mediocre predictions.’ 
• US: Veteran suicide prevention algorithm favors men, investigation finds: Military Times reports that an algorithm used by the US Department of Veteran’s Affairs to provide targeted assistance to veterans at ‘higher statistical risk of suicide’ is biased against women. The algorithm does not take into ‘military sexual trauma and intimate partner violence’ which are linked to elevated suicide risk among female veterans, and prioritizes veterans who are ‘divorced and male’ and ‘widowed and male’ for additional assistance.  
• Report: Artificial intelligence and democracy: UNESCO has released a report titled ‘Artificial intelligence and democracy’ which thematically analyzes democratic expectations and disappointments of digitization, new digital public space, the politics of Big Data, and the study of democracy as a form of political decision-making. It notes that ‘the regulation and governance of AI must be complemented by institutions facilitating public legitimization’ which is necessary for the enforcement of principles such as transparency, explainability and contestability. 

Regulation and Enforcement

• EU: EU’s AI Act to Enter Into Force Next Month: In this piece, FindBiometrics reports that EU’s AI act will come into force next month, following the endorsement from EU countries this Tuesday. 

Facial recognition, biometric identification, surveillance

• International: Researchers navigate facial recognition algorithm training with synthetic data: Biometric Update reports on a new methods developed by researchers to use synthetic data sets to train facial recognition algorithms as a way of addressing ethical and privacy concerns associated with collecting biometrics from people. 
• Hong Kong: Hong Kong slams Worldcoin for violating privacy laws with excessive facial recognition scans: This piece by Firstpost reports that Hong Kong’s Privacy Commissioner for Personal Data (PCPD) has issued an enforcement notice to Worldcoin over unnecessary and excessive collection of biometric data. The privacy watchdog has directed the company to cease biometric collection, and warned it of a fine and imprisonment on failure to comply. 
• US: These cities bar facial recognition tech. Police still found ways to access it:  The Washington Times reports that despite a ban on use of facial recognition softwares for law enforcement in many cities in the US, police departments in San Francisco and Austin have found a workaround by forwarding requests to neighboring states where such bans don’t exist. 

Other items

• The Forum on the Arms Trade will be hosting a crucial and timely event on 6 June on ‘The Ongoing Humanitarian Disarmament Agenda’: Experts will cover a number of areas including ‘The threats posed by autonomous weapons and the recent UNGA resolution’. Registration is available here!