Generative AI might hold enormous promise, but its risks have also been flagged up. These include (1) sophisticated disinformation (eg deep fakes or fake news) that could manipulate public opinion, (2) intentional exploitation of minorities and vulnerable groups, (3) historical and other biases in the data used to train generative AI models that replicate stereotypes and could lead to output such as hate speech, (4) encouraging the user to perform harmful or self-harming activities, (5) job losses in certain sectors where AI could replace humans, (6) ‘hallucinations’ or false replies, which generative AI can articulate very convincingly, (7) huge computing demands and high energy use, (8) misuse by organised crime or terrorist groups, and finally, (9) the use of copyrighted content as training data without payment of royalties.

To address those potential harms, it will be necessary to come to terms with the foundation models that underlie generative AI. Foundation models, or models through which machines learn from data, are typically trained on vast quantities of unlabelled data, from which they infer patterns without human supervision. This unsupervised learning enables foundation models to exhibit capabilities beyond those originally envisioned by their developers (often referred to as ‘emergent capabilities’).

The evolving AI Act

The proposed AI Act (European Commission, 2021), which at time of writing is still to be finalised between the EU institutions, is a poor fit for foundation models. It is structured around the idea that each AI application can be allocated to a risk category based on its intended use. This structure largely reflects traditional EU product liability legislation, in which a product has a single, well-defined purpose. Foundation models however can easily be customised to a great many potential uses, each of which has its own risk characteristics.

In the ongoing legislative work to amend the text, the European Parliament has proposed that providers of foundation models perform basic due diligence on their offerings. In particular, this should include:

• Risk identification. Even though it is not possible to identify in advance all potential use cases of a foundation model, providers are typically aware of certain vectors of risk. OpenAI knew, for instance, that the training dataset for GPT-4 featured certain language biases because over 60 percent of all websites are in English. The European Parliament would make it mandatory to identify and mitigate reasonably foreseeable risks, in this case inaccuracy and discrimination, with the support of independent experts.
• Testing. Providers should seek to ensure that foundation models achieve appropriate levels of performance, predictability, interpretability, safety and cybersecurity. Since the foundation model functions as a building block for many downstream AI systems, it should meet certain minimum standards.
• Documentation. Providers of foundation models would be required to provide substantial documentation and intelligible usage instructions. This is essential not only to help downstream AI system providers better understand what exactly they are refining or fine-tuning, but also to enable them to comply with any regulatory requirements.

Room for improvement

These new obligations, if adopted in the final AI Act, would be positive steps, but lack detail and clarity, and would consequently rely heavily on harmonised standards, benchmarking and guidelines from the European Commission. They also risk being excessively burdensome. A number of further modifications could be put in place.

Risk-based approach

Applying all obligations to the full extent to every foundation model provider, both large and small, is unnecessary. It might impede innovation and would consolidate the market dominance of firms that already have a considerable lead in FMs, including OpenAI, Anthropic and Google Deepmind. Even without additional regulatory burdens, it might be very hard for any companies outside of this group to match the resources and catch up with the FM market leaders.

A distinction could therefore be made between systemically important and non-systemically important FMs, with significantly lower burdens for the latter. This would be in line with the approach taken by the EU Digital Services Act (DSA), which notes that “it is important that the due diligence obligations are adapted to the type, size and nature of the … service concerned.” The DSA imposes much more stringent obligations on certain service providers than on others, notably by singling out very large online platforms (VLOPs) and very large online search engines (VLOEs).

There are two reasons for differentiating between systemic and non-systemic foundation models and only imposing the full weight of mandatory obligations on the former. First, the firms developing systemic foundation models (SFMs) will tend to be larger, and better able to afford the cost of intense regulatory compliance. Second, the damage caused by any deviation by a small firm with a small number of customers will tend to be far less than that potentially caused by an SFM.

There are useful hints in the literature (Bommasani et al, 2023; Zenner, 2023) as to criteria that might be used to identify SFMs, such as the data sources used, or the computing resources required to initially train the model. These will be known in advance, as will the amount of money invested in the FM. These pre-market parameters presumably correlate somewhat with the future systemic importance of a particular FM and will likely also correlate with the ability of the provider to invest in regulatory compliance. The degree to which an FM provider employs techniques that facilitate third-party access to their foundation models and thus independent verification, such as the use of open APIs, or open source, or (especially for firms that do not publish their source code) review of the code by independent, vetted experts, might also be taken into account. Other, post-deployment parameters, including the number of downloads, or use in downstream services or revenues, can only be identified after the product has established itself in the market.

Lesser burdens

Notwithstanding the arguments for a risk-based approach, even small firms might produce FMs that work their way into applications and products that reflect high-risk uses of AI. The principles of risk identification, testing and documentation should therefore apply to all FM providers, including non-systemic foundation models, but the rigour of testing and verification should be different.

Guidance, perhaps from the European Commission, could identify what these reduced testing and verification procedures should be for firms that develop non-systemic foundation models. Obligations for testing, analysis, review and independent verification could be much less burdensome and intensive (but not less than reasonably stringent) for providers of non-systemic FMs.

This kind of differentiation would allow for a more gradual and dynamic regulatory approach to foundation models. The list of SFMs could be adjusted as the market develops. The Commission could also remove models from the list if they no longer qualify as SFMs.

Use of data subject to copyright

Even though the 2019 EU Copyright Directive provides an exception from copyright for text and data mining (Article 4(1) of Directive 2019/790), which would appear in principle to permit the use of copyrighted material for training of FMs, this provision does not appear in practice to have resolved the issue. The AI Act should amend the Copyright Directive to clarify the permitted uses of copyrighted content for training FMs, and the conditions under which royalties must be paid.

Third-party oversight

The question of third-party oversight is tricky for the regulation of FMs. Is an internal quality management system sufficient? Or do increasingly capable foundation models pose such a great systemic risk that pre-market auditing and post-deployment evaluations by external experts are necessary (with protection for trade secrets)?

Given the scarcity of experts, it will be important to leverage the work of researchers and civil society to identify risks and ensure conformity. A mandatory SFM incident reporting procedure that could draw on an AI incident reporting framework under development at the Organisation for Economic Co-operation and Development might be a good alternative.

Internationally agreed frameworks

Internationally agreed frameworks, technical standards and benchmarks will be needed to identify SFMs. They could also help document their environmental impacts.

Until now, the development of large-scale FMs has demanded enormous amounts of electricity and has the potential to create a large carbon footprint (depending on how the energy is sourced). Common indicators would allow for comparability, helping improve energy efficiency throughout the lifecycle of an SFM.

Safety and security

Providers of SFMs should be obliged to invest heavily in safety and security. Cyberattacks on cutting-edge AI research laboratories pose a major risk; nonetheless, and despite rapidly growing investments in SFMs, the funding for research in AI guardrails and AI alignment is still rather low. The internal safety of SFMs is crucial to prevent harmful outputs. External security is essential, but it alone will not be sufficient – the possibility of bribes in return for access to models should be reduced as much as possible.

Conclusion

The EU is likely to be a major deployer of generative AI. This market power may help ensure that the technology evolves in ways that accord with EU values.

The AI Act is potentially ground-breaking but more precision is needed to manage the risks of FMs while not impeding innovation by smaller competitors, especially those in the EU. Unless these issues are taken into account in the finalisation of the AI Act, there is a risk of significantly handicapping the EU’s own AI developers while failing to install the adequate safeguards.

To read the full analysis by Bruegel, please click here.

J. Scott Marcus is a Senior Fellow at Bruegel, a Brussels-based economics think tank, and also works as an independent consultant dealing with policy and regulatory policy regarding electronic communications. His work is interdisciplinary and entails economics, political science / public administration, policy analysis, and engineering.

The post Adapting the European Union AI Act to Deal with Generative Artificial Intelligence appeared first on WITA.

The agreed non-agreement between the three states pertains to some of the most complex machinery and most miniscule components humankind has ever produced. With their accord, the countries effectively restricted the export to China of the most advanced microchips and the tools to produce them. These items have become a focal point in international power politics because of their use in developing artificial intelligence and their centrality to many of the 21st century’s most important technologies.

As news on the matter emerged, the Dutch prime minister confined his remarks to saying: “Those talks have been going on for a long time and we’re not saying anything about it.” The reason for reticence was clear; in response to their decision, China threatened retaliation against the Netherlands and Japan.

The move followed on from measures unilaterally implemented by the US in October 2022 to restrict the trade of advanced semiconductor technologies with China for reasons of international security. And it now appears that the Dutch national measures could soon be followed by a decision by the German government to restrict the export to China of chemicals needed for chip production.

As these sorts of incidents mount amid the escalating US-China strategic technology competition, the EU and its member states will find themselves increasingly caught in the crossfire. Washington will maintain pressure on its allies to align with its China policy. China’s military build-up will continue to change the balance of power. And Beijing’s willingness and ability to weaponise trade will likely continue to grow – it will no longer be possible for the EU to keep its pursuit of free trade separate from these powerful currents. If a rules-based order is to remain, the rules will need to change to take account of the ways in which economic security forms part of this wider competition.

To steer a course according to its own interests in this new era of strategic trade controls, the EU must urgently develop its own strategy and upgrade its tools to deliver on it. If it is to promote and defend its own sovereignty, it must start to draw its own red lines in technology engagement with China and upgrade its export control policy.

Tobias Gehrke is a senior policy fellow at the European Council on Foreign Relations, based in the Berlin office. He leads ECFR’s Geoeconomics Initiative. His area of focus includes economic security, European economic strategy, and great power competition in the global economy.

Julian Ringhof is a policy fellow with the European Power programme at the European Council on Foreign Relations. His research focuses on the implications of digital and emerging technologies for international affairs, including the topics of EU digital diplomacy and EU technological sovereignty.

To read the full policy brief, please click here.

The post The Power of Control: How the EU Can Shape the New Era of Strategic Export Restrictions appeared first on WITA.

ALI’s report focuses on next steps to creating a U.S. led global digital governance agenda. As the longer-term process of negotiating a multilateral digital agreement under the World Trade Organization evolves, the U.S. should focus on nearer-term goals in the Pacific and Europe.

A new digital agenda starts with the need to identify policies that are worker-centric. The Administration and Congress are working on a new trade agreement model to put workers at the center, and this focus needs to be part of digital agreements. This includes language covering digital inclusion and access to technology, especially to underserved communities, a focus on small- and medium-sized enterprises (SMEs) and protections for online users.

Second, the U.S. should negotiate a Pacific Digital Agreement to reestablish U.S. engagement in Asia, building on existing regional agreements, which include open and democratic values. This agreement should include a group of five or six key countries in the region, incorporate new worker-centric language, together with existing high standard language from DEPA, DEA and the U.S.-Japan Agreement, and create new norms on ethical AI, facial recognition, and technologies of the future.

Finally, the U.S. should build a coalition of like-minded, technology-democracies to develop a high standard digital governance agenda advancing open and democratic values. The U.S.-EU Tech and Trade Council is a good first step toward this goal. Building this coalition is the most critical element in countering China’s harmful approaches to tech and data governance, and the U.S. has no stronger partner in these values than the EU. However, the two sides will also need to work through digital policy friction, including privacy, taxation, and regulatory approaches like the Digital Markets Act (DMA).

ALI-White_Paper-081121-Final2

Dr. Orit Frenkel is the CEO and co-founder of the American Leadership Initiative. She has 39 years of experience working on Asia, trade, and foreign policy issues. 

Ms. Rebecca Karnak is Director of Digital Projects at the American Leadership Initiative. She is also the Principal and Founder of Woodside Policy LLC.

To read the full report from the American Leadership Initiative, please click here. 

The post Next Steps for U.S. Digital Leadership: Advancing Digital Governance with the Pacific and Europe appeared first on WITA.

Technological innovation has had a profound impact on the way we live, communicate, and work. The dawn of the Fourth Industrial Revolution has opened immense opportunities but also created significant challenges. Questions about cybersecurity, disinformation, and privacy, for example, vex businesses, governments, and private citizens alike. A different set of issues are related to the sheer size, reach, and power of the companies that comprise Big Tech and how to deal with them.

Being a large corporation, and being in the vanguard of a far-reaching and ever-expanding industry, is, by itself, neither good nor bad, but it will often lead to increased scrutiny. In some instances, this might result in attempts to either block certain companies from entering a market, or, alternatively, make it more difficult for them to operate in it. In 2015, for example, President Obama alluded to this when he accused the European Union of digital protectionism in its investigations of American tech companies— “[i]n defense of Google and Facebook, sometimes the European response … is more commercially driven than anything else.” But to chalk scrutiny of large tech companies and their business practices up to mere protectionism would miss the mark. The many benefits of modern technology notwithstanding, there are powerful economic factors within digital markets that limit competition and stifle innovation, and as a result can hurt consumers.

Concerns about Big Tech are also not confined to Europe. In fact, there seems to be a growing consensus in both the United States and the European Union of the need to, at a minimum, explore ways to check certain actions and the broader influence of the largest tech companies.

To be sure, there are differences in how Big Tech is viewed in the United States and Europe. At a basic level, many Europeans are viscerally suspicious of the market and the power of big corporations. This clearly also applies to the tech sector, as evidenced by a poll conducted in the run-up to the European Parliament elections last year. Fully 64 percent of voters thought that the European Union had been too lax in its regulation of U.S. tech giants. By contrast, most Americans believe in the power of the market to self-correct and are warier of government overreach. Whether consciously or not, it is hardly a stretch to assume that these different attitudes inform thinking about competition policy and enforcement decisions on both sides of the Atlantic.

The focus of this article is on single-firm conduct, and the transatlantic divide over how best to use antitrust and competition policy to navigate this new and exciting world. Section 2 looks at what makes Big Tech unique from an antitrust perspective. Section 3 provides an overview of U.S. and EU competition law as it relates to single-firm conduct, as well as their respective institutional structures. Section 4 assumes a more prospective posture, looking at possible future trends and what steps Big Tech can take to protect its own interests in this environment.

To read the full article please click here.

©2020 by the American Bar Association.  Reprinted with permission.  All rights reserved.  This information or any or portion thereof may not be copied or disseminated in any form or by any means or stored in an electronic database or retrieval system without the express written consent of the American Bar Association.

Use of the Requested Material is limited to the one-time use as described above, and does not include the right to license this Requested Material, individually or as it appears in your publication, or to grant others permission to photocopy or otherwise reproduce this material.  Permission is granted to make versions for use by blind or physically handicapped persons, provided that no fees are charged.

Use of the Requested Material is granted on a non-exclusive basis and is valid throughout the world in the English language only.

If any material in the Requested Material credits another source, then you must obtain authorization from that original source.

Permission is limited solely to the text portion of the Requested Material. If any photographs, illustrations, cartoons, advertisements, etc. appear in conjunction with the Requested Material, those portions should be blocked out before reproduction, as well as text from other articles.

The reproduction of the ABA logo and/or section logos is strictly prohibited, as is the reproduction of covers and mastheads of ABA publications

The post Antitrust in the United States and the European Union – A Comparative Analysis appeared first on WITA.

This is all the more necessary that the 2019 elections have resulted in a rebalancing of political forces at the European Parliament which will necessitate to review the trade environment nexus at EU level for several reasons:

• environment protection featured prominently among the political signals sent by the voters;
• it is, by essence, a global public good issue, better dealt with at EU level;
• the EU is seen as having so far exercised a leadership role in this area of global governance;
• trade is one of the few really “federalised” EU competences;
• as such, it remains the main EU lever to influence the global agenda, starting with SDGs.

This is confirmed by noticeable developments since the elections, such as the new President of the Commission declaring that she is in favour of border carbon taxes (a first), or by the growing debate on the preservation of the rainforest that have surfaced as a result of the EU and Mercosur’s agreement reached after 25 years of bilateral trade negotiations.

Even if trade measures are not among the “first best solutions” to tackle environmental degradations, revisiting the EU stance in this area appears, both necessary and urgent, starting with climate change related aspects. This is also true about other issues such as biodiversity or ocean governance. It is a highly complex matter, necessitating deep analytical and technical investigations in several areas, new political debates, and search for operational
and implementable solutions.

To download the full paper, please click here.

The post Time To Green EU Trade Policy: But How? appeared first on WITA.

Europe’s current and future sustainable economic growth and societal wellbeing increasingly draws on value created by data. AI is one of the most important applications of the data economy. Today most data are related to consumers and are stored and processed on central cloud-based infrastructure. By contrast a large share of tomorrow’s far more abundant data will come from industry, business and the public sector, and will be stored on a variety of systems, notably on computing devices working at the edge of the network.

This opens up new opportunities for Europe, which has a strong position in digitised industry and business-to-business applications, but a relatively weak position in consumer platforms. Simply put, AI is a collection of technologies that combine data, algorithms and computing power. Advances in computing and the increasing availability of data are therefore key drivers of the current upsurge of AI.

The post White Paper on Artificial Intelligence: a European approach to excellence and trust appeared first on WITA.