The AI Act, the continent’s comprehensive regulation governing the use of AI technologies and their impact on citizen’s rights, came into force in early August, but will take around two years for member states to fully implement it.
“We are proud to sign onto the EU AI Pact with other industry leaders and to reinforce our commitment to AI governance by ensuring that our systems meet the highest standards of trust, transparency, and accountability,” said Eric Loeb, executive VP for government affairs at Salesforce.
“We welcome the opportunity to collaborate with industry partners, policymakers, and the broader AI community to build a future where AI is both trusted and transformative.”
Subscribe today for free
The EU AI Act categorises AI systems deployed throughout the continent against the level of risk they pose to a citizen’s freedom, ranging from low to high.
Businesses deploying AI systems deemed high risk would be subject to strict obligations, including mandatory risk assessments and log maintenance of its use as well as adequate oversights.
Some high-risk AI — such as biometric categorisation systems or emotional recognition systems — would be outright banned.
Enterprises founded to be using AI systems that violate the Act face hefty fines, ranging from $8 million or 1.5% of global annual revenue up to $38 million or 7% of revenue.
AI Pact commitments
Given the AI Act is some two years away, the European Commission launched the AI Pact to encourage businesses using AI across the continent to get an early start with compliance.
All participants of the EU AI Pact agree to three key pledges:
AI governance strategy to foster the uptake of AI and commitments to future compliance with the AI Act.
Identifying AI systems likely to be categorised as high-risk under the AI Act
Promoting AI literacy and awareness among staff, ensuring ethical and responsible AI development.
Beyond the core commitments, the Commission said that more than half of the signatories agreed to take on additional pledges, including labelling certain types of AI-generated content and ensuring human oversight.
Signatories include multinational companies and European small and medium enterprises (SMEs) from sectors spanning IT, telecoms, and healthcare, as well as banking, automotive, and aeronautics.
Notable names appearing on the signatory list include companies developing AI models, such as OpenAI, Adobe, Cohere, and IBM. Also found on the list are tech giants including Google, Microsoft, Qualcomm, Samsung, and SAP.
Among the telcos signed on include Telefónica, Vodafone, Nokia, Deutsche Telekom, TIM Telecom Italia, and Telenor Nordics.
Infrastructure players like Hewlett Packard Enterprise, Cisco, and Palo Alto Networks have also pledged.
A notable absentee from the EU AI Pact is Meta, the parent company of social media sites like Facebook and Instagram and the maker of the popular Llama open source AI models.
Meta has publicly been at odds with the AI Act as the company claims the “fragmented” regulation prevents it from using European-originated data to train its AI models and systems.
Meta, along with Ericsson, Spotify, and SAP recently called for the regulation to be amended to provide greater clarity around data usage.
The companies claimed the EU would lose out to the US and China with AI systems on the continent being “less competitive and less innovative” if they’re not able to use European-originated data as part of their datasets.
RELATED STORIES
Meta, Ericsson, SAP warn EU AI Act may hinder innovation and competitiveness
EU market chief Thierry Breton resigns amid clash with von der Leyen
Industry leaders echo Draghi report: EU telecoms needs investment, consolidation
