This browser is not actively supported anymore. For the best passle experience, we strongly recommend you upgrade your browser.
| 4 minutes read

Encouraging planning ahead: The European Commission's AI Pact

The EU’s AI Act is set to impose new obligations on both developers and deployers of AI systems, obligations which will be significant with respect to “high-risk” AI models. The substantive obligations will be phased in over a three-year period, but more than 100 companies including Microsoft, Google, Amazon and OpenAI have signed up to the EU Commission’s AI Pact and have committed to early implementation of the AI Act's requirements. 

The AI Pact encourages and supports organisations to plan ahead for the implementation of AI Act measures. Signatories have agreed to make best efforts to meet the Pact’s core commitments, which focus on transparency obligations and requirements for high-risk AI systems, and will engage with other signatories to gather vital insights on the implementation of AI Act.  We take a look at the pros and cons.

What does the AI Pact do?

The AI Pact requires a voluntary commitment from its signatories to implement the AI Act’s requirements ahead of legal deadlines and to report on their compliance regularly. This involves a comprehensive approach to organisational governance, encompassing 

  • the establishment of risk management systems
  • data governance practices
  • technical documentation
  • automatic record-keeping, and 
  • maintaining transparency and human oversight in AI operations. 

Specifically, organisations must develop an AI governance strategy, map out AI systems considered high-risk, promote AI literacy among staff, and comply with high-risk AI system obligations.

The two pillars’ approach

The Commission’s AI Pact rests on two pillars: 

  • Pillar I: Revolving around the network of participants to the AI Pact, it supports the creation of an agora where organisations can share their experiences and best practices, and gain knowledge from other members. Participants will have the opportunity to take part in workshops organised by the AI Office, which will also keep track of how companies are implementing the AI Act rules and any challenges faced. 
     
  • Pillar II:  Focusing on early implementation of the AI Act’s measures, participants are encouraged to disclose the steps they are taking and measures they are putting in place to comply with the new rules, including concrete actions and adoption timelines.

A diagram of a diagram

Description automatically generated

Source: AI Pact | Shaping Europe’s digital future (europa.eu)

The pros and cons of signing up…

Signing up to the AI Pact may bring several notable advantages:

  • Head start: The AI Pact would allow organisations to achieve early compliance and preparedness for the AI Act. This proactive stance provides a valuable head start in adapting to forthcoming regulations, helping to avoid the last-minute rush and potential penalties for non-compliance. 
     
  • Building trust: Committing to the AI Pact can significantly enhance an organisation's credibility and trustworthiness. By publicly demonstrating a commitment to ethical and responsible AI usage, companies can build trust among customers, investors, and regulators, fostering a positive reputation. 
     
  • Gaining practical insights: The AI Pact facilitates knowledge sharing and best practices among its members through workshops and an online platform, promoting continuous improvement in strategies and operations. Participation in the AI Pact offers substantial visibility and networking opportunities within the broader AI community, thereby enhancing engagement and collaboration.

On the flip side, there are potential downsides to consider:

  • Substantial upfront investment: The primary disadvantage is the resource-intensive nature of meeting the AI Pact's requirements ahead of legal deadlines. Complying early necessitates significant investments of time, money, and effort before it is legally required. This early investment can be substantial, potentially causing operational disruptions as companies work to adapt to new policies and practices. 
     
  • Reputational impact: There are reputational considerations for companies to bear in mind. Signatories may attract greater attention and scrutiny, as committing early positions them as leaders in ethical AI practices. In such a scenario, there is a risk of being found not fully compliant, which can impact their reputation, even though they cannot be formally sanctioned before the AI Act actually kicks in. Conversely, if the AI Pact gains significant momentum, companies that choose not to subscribe may suffer.

…and its cost 

Complying with the AI Pact is likely come with a financial burden for signatories, with costs related to compliance efforts, policy development and the implementation of new business practices: 

  • Signatories will need to ensure AI systems comply with the complex requirements of the AI Act and specific commitments made under the AI Pact, which will involve ongoing consultations and legal oversight. 
     
  • Policy teams must develop, implement, and monitor new organisational policies to meet governance, risk management, and transparency obligations. 
     
  • Signatories might need to establish or enhance internal processes, such as data governance frameworks, technical infrastructure for automatic record-keeping, and systems for human oversight. 

This extensive compliance effort involves training staff, refining business operations, and potentially investing in new technologies and systems. The financial outlay may vary significantly based on the size of the organisation and the complexity of its AI operations, but it will include both initial setup costs and ongoing operational expenses to maintain compliance.

Looking ahead

The AI Pact does not impose legally binding obligations on, but signatories will be asked to report on the outcome of the implementation 12 months after the publication of their commitments. 

The Commission has left the door open for other companies to join the AI Pact until the AI Act becomes fully applicable. It has been reported that around 1000 companies have expressed interest in joining the Pact, so it will be interesting to see whether take-up continues and how useful organisations find it as a means for exchanging information and working out how best to achieve compliance.

Those interested in joining the AI Pact initiative can sign up here: EUSurvey - Survey (europa.eu)

The AI Pact encourages and supports organisations to plan ahead for the implementation of AI Act measures.

Tags

ai