AI

UK’s competition watchdog drafts principles for ‘responsible’ generative AI

Comment

Digital image of artificial intelligence human brain on black background.
Image Credits: Andriy Onufriyenko (opens in a new window) / Getty Images

An initial review of generative AI by the U.K.’s Competition and Markets Authority (CMA) which was announced back in May has concluded with a report containing seven proposed principles to “ensure consumer protection and healthy competition are at the heart of responsible development and use of foundation models” (FMs), as it puts it.

The principles the competition watchdog has come up with for consideration, as it kicks off another round of stakeholder engagement on AI’s potential impacts on markets, are:

  • Accountability: “FM developers and deployers are accountable for outputs provided to consumers”
  • Access: “ongoing ready access to key inputs, without unnecessary restrictions”
  • Diversity: “sustained diversity of business models, including both open and closed”
  • Choice: “sufficient choice for businesses so they can decide how to use FMs”
  • Flexibility: “having the flexibility to switch and/or use multiple FMs according to need”
  • Fair dealing: “no anti-competitive conduct including anti-competitive self-preferencing, tying or bundling”
  • Transparency: “consumers and businesses are given information about the risks and limitations of FM-generated content so they can make informed choices”

The competition watchdog is drawing on its experience regulating market contestibility combined with some early research and feedback from AI stakeholders to pull together this first draft of pro-innovation principles. The move follows instruction from the U.K. government to existing regulators to consider AI impacts on their patches. Down the line, the CMA might thus end up promoting such a list as best practice for avoiding competition complaints at AI’s cutting edge.

Nothing is set in stone yet, though — with another update on its thinking in this area planned for early 2024. So watch this space.

Scoping AI impacts

“In this market for foundation models, there’s lots at stake for both competition and consumers. If the market works well, the best products win. And so do consumers and so do people. But if it doesn’t, people could really lose out and compelling businesses could struggle to compete. So with this review… we wanted to be on the on the front foot as much as possible — trying to understand what’s going on as it happens, rather than having to come in later and figure it out after the fact,” said Will Hayter, senior director for the CMA’s Digital Markets Unit (DMU), speaking in an interview with TechCrunch.

“There are tonnes of potential benefits for these models. But also, of course, some risks. And we think the benefits, or the harms, could happen quite fast. So we’ve tried to focus on some potential positive outcomes and some potential less positive outcomes. And then we’ve really thought hard about the kinds of drivers that might push in one direction or another.”

By foundation models the CMA said its focus is on large-scale AI models that can be adapted to downstream customer applications through a process of fine-tuning — so which play a particular role in the AI supply chain where they’re intended to be built on by others developing customer facing apps and services.

Hayter confirmed it’s too soon for the CMA to have an established view on how these still fast-developing AI technologies might impact markets. Nor could he say whether foundation model makers could be future candidates for bespoke regulation under the U.K.’s planned “pro-competition” reform of rules applied to Big Tech with so-called “Strategic Markets Status” (a long-trailed digital regulation reboot which was recently revived by prime minister Rishi Sunak) — telling us “I think it really would be wrong to prejudge and try and forecast that at this point”.

But the regulator is clearly keen to be proactive in scoping out a flavor of cutting-edge tech with such major potential for impact.

“We think it’s important to try and nudge the market towards some of those more positive outcomes. And that’s what the set of proposed principles are trying to achieve,” said Hayter, adding: “But I really would emphasise they’re genuinely proposed at this stage.

“We produced the report but that means we can now get out and discuss both the content but also the principles with lots of different types of organisation… to see how those principles might be improved upon and also how they might be adopted to try and get to those more positive outcomes. So we’re pretty much at the beginning of the conversation, which we’re looking forward to having.”

OpenAI, DeepMind and Anthropic to give UK early access to foundational models for AI safety research

The CMA received some 70+ responses following its call for input ahead of the review. Hayter would not be drawn into a breakdown of where this feedback came from but he suggested they’d heard from “a broad range of types of organisations — from the [AI] labs themselves, the big firms, some civil society, academics, a range of experts”, as well as conducting their own research to feed the report.

“We’ve pulled in quite a lot of varied inputs. But, again, what I’m really looking forward to now is now that we have our initial thoughts out there, and these principles, we can really use that as a bit of a framework for having the next round of conversations — and see how we might be able to work together, as collaborative as you possibly can with a wide range of types of organisations, to try and help get the market to the best possible place,” he added.

The U.K. government came up with its own set of principles to guide the development of AI back in March in its policy white paper. While these two lists have some overlap (the government’s five principles for AI are: safety, security and robustness; transparency and explainabilityfairness; accountability and governance; contestability and redress) the CMA’s proposed principles are specifically targeted at potential risks that fall within its competition and consumer protection purview.

It’s also worth noting it has also not looked at a full spectrum of potential consumer concerns — for example it notes that issues like security and data protection are not considered in the initial review. Here it appears keen to ensure it stays well within its regulatory lane (whereas issues like data security and privacy fall more obviously under the Information Commissioner’s Office — which is also issuing guidance for generative AI developers).

Asked about this potential gap the CMA emphasized it will be working with other U.K. regulators also tasked with paying attention to AI under the government’s plan for developing context-specific guidance — pointing to the Digital Regulatory Cooperation Forum (which was established by the CMA, ICO and Ofcom in July 2020, with the FCA joining as a full member in April 2021) as playing an important role in any relevant joint working.

Targeted approach

One wider question competition regulators may face is how to balance the need to let novel AI technologies (and business models) develop vs responding to a sense of urgency in light of the tech’s scale and potency. And, well, the need for speed to tackle what could be a new wave of problems being baked into digital markets that have already — for years — been characterized by issues like tipping and unfair dealing, while consumers have also faced exploitative business models foisted on them by dominant platforms operating under their own self-serving T&Cs.

Such concerns lie behind the U.K. plan to reboot the domestic competition regime by adding a proactive regime of bespoke rules that can be applied to the most powerful platforms by the DMU, where Haytor is a senior director.

The European Union has its own flavor of ex ante digital competition reform already — which is up and running (aka the Digital Markets Act, which is applicable to Alphabet, Amazon, Apple, ByteDance, Meta and Microsoft). While Germany has, since a few years, implemented its own update to competition rules targeted at Big Tech. So the U.K. is lagging behind peers in addressing tech giants’ market power.

On foundation models, Hayter suggested some (possible) negative scenarios vis-a-vis market impacts could flow from similar issues to those that have entrenched (current gen) Big Tech. However he said there’s still too much uncertainty about how AI power plays will go to predict whether markets will track towards another wave of AI-fuelled concentration (i.e. powered by a few dominant foundation model makers) or blossom into the opposite: Vibrant competition as businesses tap into the power of FMs.

Both scenarios are possible in his view.

“You could see scenarios where these models actually help newer challengers challenge big incumbent positions and that’ll be great… That could challenge those positions of market power,” he argued. “On the flip side, of course, depending on some of these issues that we’ve highlighted — things like how the access to key inputs might be controlled — you could get the opposite situation where actually these foundation models are a help to firms in currently strong positions to shore up those positions further.

“That will all really depend on the specific context and specific market — and you might see one scenario isn’t and then the other the opposite in another — so we need to we need to be really, really focused on the evidence and on the specifics in the market and  be prepared to act when necessary but not also not jump in too quickly.”

Despite what could be seen as an early intervention by the CMA to get a handle on emerging AI developments, Hayter’s top-line message is one that’s likely to reassure industry: The U.K. regulator won’t be rushing to rein in the cutting edge.

Any future rules (or even a set of confirmed principles) would need to be “very closely targeted”, he emphasized.

“We certainly need to be very measured, and I think whatever kind of regulation… whether it’s for competition or other reasons, that would need to be very closely targeted to specific questions and issues/problems based on real evidence,” he said, adding: “This report is certainly not suggesting jumping in and regulating… It’s trying to identify the kinds of things that would help realise the maximum potential from the technology and also to be aware of things to look out for.”

“We’re trying extremely hard not to prejudge what’s happening here,” he went on. “We’re trying to work out… what the specific drivers are that might push in one direction or another. So we do highlight access to data. And the driver that we explored in that area was whether access to proprietary data will become important. We have heard that at the moment, there’s a reasonable availability of publicly available data to train models on but it’s possible that, over time, the role of proprietary data becomes bigger — and then, as you say, that might play into the hands of companies that have large banks and proprietary data. But we don’t think that’s happened at this point. And it still might not happen — that depends on how the market develops. And similarly on the access to compute.

“That’s clearly an extremely important input to these models, which is why we highlighted access as one of these key principles. And… there’s work going on elsewhere to consider access to public cloud services in the work that Ofcom is doing and has previously proposed to send to us as market investigation. So we waited to see the outcome of that.”

“There’s a broader question about the ability of the current frameworks to respond to new developments,” he also told us during our interview. “So you will have seen us in a number of contexts explain that the existing tools — i.e. competition, and enforcement of Competition and Consumer Law — sometimes can be a bit slow to respond to some of the specific developments. And that’s what the progress of the Digital Markets Competition and Consumers bill through parliament is intended to help with. But just to emphasise that again… is all about being very targeted at very specific problems and with an intentionally high hurdle for taking any action, which is the concept of Strategic Market Status in that framework.

“The Digital Markets Competition and Consumers bill — the Strategic Market Status framework — gives us the broad structure to be able to tackle some of these problems that arise in digital markets. But we certainly wouldn’t want to rush in too quickly and do anything here [with FMs] because we still think this market could develop in more positive directions, particularly if these principles are backed up.”

UK’s antitrust watchdog announces initial review of generative AI

EU to let ‘responsible’ AI startups train models on its supercomputers

More TechCrunch

Ahead of the AI safety summit kicking off in Seoul, South Korea later this week, its co-host the United Kingdom is expanding its own efforts in the field. The AI…

UK opens office in San Francisco to tackle AI risk

Companies are always looking for an edge, and searching for ways to encourage their employees to innovate. One way to do that is by running an internal hackathon around a…

Why companies are turning to internal hackathons

Featured Article

I’m rooting for Melinda French Gates to fix tech’s broken ‘brilliant jerk’ culture

Women in tech still face a shocking level of mistreatment at work. Melinda French Gates is one of the few working to change that.

9 hours ago
I’m rooting for Melinda French Gates to fix tech’s  broken ‘brilliant jerk’ culture

Blue Origin has successfully completed its NS-25 mission, resuming crewed flights for the first time in nearly two years. The mission brought six tourist crew members to the edge of…

Blue Origin successfully launches its first crewed mission since 2022

Creative Artists Agency (CAA), one of the top entertainment and sports talent agencies, is hoping to be at the forefront of AI protection services for celebrities in Hollywood. With many…

Hollywood agency CAA aims to help stars manage their own AI likenesses

Expedia says Rathi Murthy and Sreenivas Rachamadugu, respectively its CTO and senior vice president of core services product & engineering, are no longer employed at the travel booking company. In…

Expedia says two execs dismissed after ‘violation of company policy’

Welcome back to TechCrunch’s Week in Review. This week had two major events from OpenAI and Google. OpenAI’s spring update event saw the reveal of its new model, GPT-4o, which…

OpenAI and Google lay out their competing AI visions

When Jeffrey Wang posted to X asking if anyone wanted to go in on an order of fancy-but-affordable office nap pods, he didn’t expect the post to go viral.

With AI startups booming, nap pods and Silicon Valley hustle culture are back

OpenAI’s Superalignment team, responsible for developing ways to govern and steer “superintelligent” AI systems, was promised 20% of the company’s compute resources, according to a person from that team. But…

OpenAI created a team to control ‘superintelligent’ AI — then let it wither, source says

A new crop of early-stage startups — along with some recent VC investments — illustrates a niche emerging in the autonomous vehicle technology sector. Unlike the companies bringing robotaxis to…

VCs and the military are fueling self-driving startups that don’t need roads

When the founders of Sagetap, Sahil Khanna and Kevin Hughes, started working at early-stage enterprise software startups, they were surprised to find that the companies they worked at were trying…

Deal Dive: Sagetap looks to bring enterprise software sales into the 21st century

Keeping up with an industry as fast-moving as AI is a tall order. So until an AI can do it for you, here’s a handy roundup of recent stories in the world…

This Week in AI: OpenAI moves away from safety

After Apple loosened its App Store guidelines to permit game emulators, the retro game emulator Delta — an app 10 years in the making — hit the top of the…

Adobe comes after indie game emulator Delta for copying its logo

Meta is once again taking on its competitors by developing a feature that borrows concepts from others — in this case, BeReal and Snapchat. The company is developing a feature…

Meta’s latest experiment borrows from BeReal’s and Snapchat’s core ideas

Welcome to Startups Weekly! We’ve been drowning in AI news this week, with Google’s I/O setting the pace. And Elon Musk rages against the machine.

Startups Weekly: It’s the dawning of the age of AI — plus,  Musk is raging against the machine

IndieBio’s Bay Area incubator is about to debut its 15th cohort of biotech startups. We took special note of a few, which were making some major, bordering on ludicrous, claims…

IndieBio’s SF incubator lineup is making some wild biotech promises

YouTube TV has announced that its multiview feature for watching four streams at once is now available on Android phones and tablets. The Android launch comes two months after YouTube…

YouTube TV’s ‘multiview’ feature is now available on Android phones and tablets

Featured Article

Two Santa Cruz students uncover security bug that could let millions do their laundry for free

CSC ServiceWorks provides laundry machines to thousands of residential homes and universities, but the company ignored requests to fix a security bug.

2 days ago
Two Santa Cruz students uncover security bug that could let millions do their laundry for free

TechCrunch Disrupt 2024 is just around the corner, and the buzz is palpable. But what if we told you there’s a chance for you to not just attend, but also…

Harness the TechCrunch Effect: Host a Side Event at Disrupt 2024

Decks are all about telling a compelling story and Goodcarbon does a good job on that front. But there’s important information missing too.

Pitch Deck Teardown: Goodcarbon’s $5.5M seed deck

Slack is making it difficult for its customers if they want the company to stop using its data for model training.

Slack under attack over sneaky AI training policy

A Texas-based company that provides health insurance and benefit plans disclosed a data breach affecting almost 2.5 million people, some of whom had their Social Security number stolen. WebTPA said…

Healthcare company WebTPA discloses breach affecting 2.5 million people

Featured Article

Microsoft dodges UK antitrust scrutiny over its Mistral AI stake

Microsoft won’t be facing antitrust scrutiny in the U.K. over its recent investment into French AI startup Mistral AI.

2 days ago
Microsoft dodges UK antitrust scrutiny over its Mistral AI stake

Ember has partnered with HSBC in the U.K. so that the bank’s business customers can access Ember’s services from their online accounts.

Embedded finance is still trendy as accounting automation startup Ember partners with HSBC UK

Kudos uses AI to figure out consumer spending habits so it can then provide more personalized financial advice, like maximizing rewards and utilizing credit effectively.

Kudos lands $10M for an AI smart wallet that picks the best credit card for purchases

The EU’s warning comes after Microsoft failed to respond to a legally binding request for information that focused on its generative AI tools.

EU warns Microsoft it could be fined billions over missing GenAI risk info

The prospects for troubled banking-as-a-service startup Synapse have gone from bad to worse this week after a United States Trustee filed an emergency motion on Wednesday.  The trustee is asking…

A US Trustee wants troubled fintech Synapse to be liquidated via Chapter 7 bankruptcy, cites ‘gross mismanagement’

U.K.-based Seraphim Space is spinning up its 13th accelerator program, with nine participating companies working on a range of tech from propulsion to in-space manufacturing and space situational awareness. The…

Seraphim’s latest space accelerator welcomes nine companies

OpenAI has reached a deal with Reddit to use the social news site’s data for training AI models. In a blog post on OpenAI’s press relations site, the company said…

OpenAI inks deal to train AI on Reddit data

X users will now be able to discover posts from new Communities that are trending directly from an Explore tab within the section.

X pushes more users to Communities