ARTIFICIAL INTELLIGENCE ACT: TECHNICAL ASSISTANCE FOR AI SAFETY

The European Commission, specifically DG CNECT (Communications Networks, Content and Technology), has issued a call for tenders, titled "ARTIFICIAL INTELLIGENCE ACT: TECHNICAL ASSISTANCE FOR AI SAFETY," identified as EC-CNECT/2025/OP/0032. This procurement opportunity aims to secure expert technical assistance for the AI Office, which supports the monitoring of compliance with the Artificial Intelligence Act (Regulation EU 2024/1689). The focus is on assessing risks posed by General-Purpose AI (GPAI) models at the Union level. The estimated total funding for this procurement is 9,330,000 EUR.

The AI Act, effective from August 1, 2024, with specific rules applying from August 2, 2025, establishes responsibilities for providers of GPAI models to ensure transparency, safety, and accountability. The AI Office is tasked with monitoring compliance and evaluating systemic risks associated with these models.

The tender is segmented into six lots, focusing on distinct systemic risks. Lot 1 pertains to CBRN (chemical, biological, radiological, and nuclear) risks, with an emphasis on improving the AI Office's capacity for risk identification and assessment, valued at 1,850,000 EUR. Lot 2 covers cyber offence risks associated with sophisticated cyber attacks, with a fund of 2,000,000 EUR. Lot 3 addresses losses of control related to autonomous models, valued at 1,800,000 EUR. Lot 4 focuses on harmful manipulation risks that may induce unwanted decisions, budgeted at 1,300,000 EUR. Lot 5 tackles sociotechnical risks stemming from bias or discrimination, also at 1,300,000 EUR. Lastly, Lot 6 involves creating an agentic evaluation interface for evaluating GPAI, valued at 1,080,000 EUR.

The tasks shared across these lots include organizing risk modeling workshops, developing new evaluation tools, integrating existing evaluations into a risk framework, and offering expert risk monitoring services. All lots involve service contracts with a maximum duration of 24 to 36 months, awarded based on the best price-quality ratio.

This is an open call for tenders requiring electronic submission, with submissions due by August 25, 2025, at 16:00 CEST. Interested participants must utilize the provided electronic systems for submission, ensuring compliance with all specified technical and procedural guidelines.

The call emphasizes the need for expertise in AI safety, risk modeling, cybersecurity, and related fields, offering an opportunity for organizations to support the implementation and enforcement of the AI Act and contribute to the development of safe AI technologies.

The European Commission, DG CNECT - Communications Networks, Content and Technology, has published a call for tenders, EC-CNECT/2025/OP/0032, titled "ARTIFICIAL INTELLIGENCE ACT: TECHNICAL ASSISTANCE FOR AI SAFETY." The tender aims to secure technical assistance for the AI Office to support the monitoring of compliance with Regulation (EU) 2024/1689, the Artificial Intelligence Act, specifically concerning the assessment of risks posed by General-Purpose Artificial Intelligence (GPAI) models at the Union level. The estimated total value of this procurement is 9,330,000 EUR. The TED reference is 130/2025 450513-2025.

The AI Act, which entered into force on August 1, 2024, with certain rules applicable from August 2, 2025, establishes responsibilities for GPAI model providers to ensure transparency, safety, and accountability. Articles 89 and 92 of the AI Act empower the AI Office to monitor compliance and evaluate GPAI models for systemic risks. Article 93 allows the Commission to request providers to take corrective measures, implement mitigation strategies, or restrict model availability.

This tender is divided into six lots. Five lots are dedicated to specific systemic risks:

Lot 1: CBRN Risk Modelling and Evaluation. This lot focuses on supporting the AI Office in enforcing the AI Act concerning GPAI models that may pose risks of enabling chemical, biological, radiological, and nuclear attacks or accidents. The AI Office aims to improve its capacity in identifying, monitoring, and assessing these risks, including measuring relevant capabilities of GPAI models and assessing the effectiveness of mitigations. The estimated value is 1,850,000 EUR.

Lot 2: Cyber Offence Risk Modelling and Evaluation. This lot addresses risks related to offensive cyber capabilities that could enable large-scale or sophisticated cyber-attacks, including on critical systems. This includes automated vulnerability discovery, exploit generation, operational use, and attack scaling. The AI Office seeks to improve its capacity in identifying, monitoring, and assessing these risks. The estimated value is 2,000,000 EUR.

Lot 3: Loss of Control Risk Modelling and Evaluation. This lot focuses on risks related to the inability to oversee and control autonomous GPAI models that may result in large-scale safety or security threats. This includes model capabilities and propensities related to autonomy, alignment with human intent or values, self-reasoning, self-replication, self-improvement, evading human oversight, deception, or resistance to goal modification. The estimated value is 1,800,000 EUR.

Lot 4: Harmful Manipulation Risk Modelling and Evaluation. This lot addresses risks of enabling the targeted distortion of the behavior of persons, in particular through multi-turn interactions, that causes them to take a decision that they would not have otherwise taken, in a manner that causes, or is reasonably likely to cause, significant harm on a large scale. This includes the capability to manipulate through multi-turn interaction and the propensity of models to manipulate, including manipulation of high-stakes decision makers, large-scale fraud, or exploitation of people based on protected characteristics. The estimated value is 1,300,000 EUR.

Lot 5: Sociotechnical Risk Modelling and Evaluation. This lot focuses on large-scale sociotechnical risks, including those stemming from harmful bias or discrimination, or from the endangering of fundamental human rights such as freedom of expression or health protection. The AI Office aims to enhance its ability to identify, monitor, and evaluate these risks. The estimated value is 1,300,000 EUR.

These five lots share similar tasks and services:

Organization of risk modeling workshops with the AI Office.
Development of new evaluation tools private to the Commission.
Onboarding of existing publicly available evaluations into the developed risk framework.
Creation of a reference procedure and reporting template for risk assessment.
Ad hoc services for complex evaluations, including red-teaming and uplift studies.
A risk monitoring service with regular briefings on new developments.

Lot 6: Agentic Evaluation Interface. This lot is cross-cutting and involves supplying the AI Office with a programmatic interface for evaluating GPAI models through agentic interaction patterns. This includes software, methodology, cloud infrastructure support, and a methodology to evaluate GPAI on diverse types of benchmarks, focusing on agentic interaction and complex agentic benchmarks. The estimated value is 1,080,000 EUR.

The nature of the contract for all lots is services. The maximum contract duration is 36 months for lots 1-5 and 24 months for lot 6. The award method is the best price-quality ratio. The lead contracting authority is the European Commission, DG CNECT.

Key milestones include a TED publication date of July 10, 2025, a public opening date of August 26, 2025, at 14:00 Europe/Brussels time, and a deadline for receipt of tenders on August 25, 2025, at 16:00 Europe/Brussels time. The contracting authority is not bound to reply to questions submitted after August 14, 2025, at 05:00 Europe/Brussels time.

The available documents include the invitation letter, tender specifications, declaration on honor, financial offer form, annexes, model service contract, and annexes related to security requirements and security baseline.

Submissions must be sent exclusively via electronic submission.

The eSubmission system supports the latest versions of Google Chrome and Mozilla Firefox. Attachments should be less than 50 MB in size, and the maximum number of attachments per submission is 200. A Participant Identification Code (PIC) is mandatory for participation. The system supports all 24 official EU languages and uses UTF-8 character set encoding. The system encrypts all uploaded documents. Submissions in draft status can be edited, viewed, and deleted until the submission deadline.

In summary, this tender is about acquiring technical assistance for the European Commission's AI Office to enforce the AI Act, particularly in evaluating the safety and risks associated with General-Purpose AI models. The tender is divided into six lots, each focusing on different aspects of AI safety, from CBRN risks and cyber offenses to loss of control, harmful manipulation, sociotechnical risks, and the development of an agentic evaluation interface. The goal is to ensure that AI technologies are developed and used in a transparent, safe, and accountable manner, in line with the EU's regulatory framework. Companies with expertise in AI risk assessment, cybersecurity, cloud infrastructure, and software development are encouraged to apply.

Eligible Applicant Types: The eligible applicant types are not explicitly stated, but based on the nature of the tender, it is likely that eligible applicants include organizations with expertise in AI safety, risk modeling, cybersecurity, CBRN, sociotechnical analysis, software development, cloud infrastructure, and related fields. This could include SMEs, large enterprises, research institutes, universities, and consulting firms.

Funding Type: The funding type is procurement, specifically a service contract.

Consortium Requirement: The tender does not explicitly state whether a single applicant or a consortium is required. However, given the complexity and breadth of the tasks, a consortium might be advantageous, although single applicants are not excluded.

Beneficiary Scope (Geographic Eligibility): The geographic eligibility is not explicitly stated, but it is implied that the opportunity is open to entities within the EU and possibly EEA, and potentially other countries depending on specific requirements outlined in the tender documents.

Target Sector: The target sector is Artificial Intelligence, specifically AI safety, risk assessment, and technical assistance related to the AI Act. Sub-sectors include cybersecurity, CBRN (chemical, biological, radiological, and nuclear) risk, sociotechnical risk, software development, and cloud infrastructure.

Mentioned Countries: No specific countries are mentioned, but the focus is on the EU due to the AI Act and the AI Office being the lead contracting authority.

Project Stage: The project stage is focused on development, validation, and demonstration, as the tender seeks technical assistance for monitoring compliance and evaluating risks associated with GPAI models, as well as developing new evaluation tools and methodologies.

Funding Amount: The estimated total value of the tender is 9,330,000 EUR. The estimated values for individual lots vary: Lot 1 (CBRN Risk Modelling and Evaluation) is 1,850,000 EUR, Lot 4 (Harmful Manipulation Risk Modelling and Evaluation) is 1,300,000 EUR, Lot 2 (Cyber Offence Risk Modelling and Evaluation) is 2,000,000 EUR, Lot 3 (Loss of Control Risk Modelling and Evaluation) is 1,800,000 EUR, Lot 5 (Sociotechnical Risk Modelling and Evaluation) is 1,300,000 EUR, and Lot 6 (Agentic Evaluation Interface) is 1,080,000 EUR.

Application Type: The application type is an open call for tenders, with electronic submission.

Nature of Support: Beneficiaries will receive money in exchange for providing services, software, cloud infrastructure support, and technical assistance.

Application Stages: The number of application stages is not explicitly stated, but it can be inferred that there will be at least two stages: submission of the tender and evaluation of the tender.

Success Rates: The success rates are not mentioned.

Co-funding Requirement: The co-funding requirement is not mentioned.

Summary:

This opportunity is a call for tenders by the European Commission, DG CNECT, for technical assistance to the AI Office in enforcing the AI Act, specifically concerning the safety and risk assessment of General-Purpose Artificial Intelligence (GPAI) models. The tender is divided into six lots, each focusing on different aspects of AI risk: CBRN risks, cyber offence risks, loss of control risks, harmful manipulation risks, sociotechnical risks, and the development of an agentic evaluation interface.

The AI Act, Regulation (EU) 2024/1689, which has entered into force on the 1st of August 2024, and for which certain rules will become applicable on the 2nd of August 2025 establishes a comprehensive legal framework governing the area of Artificial Intelligence, and in particular relevance to this procurement, establishes responsibilities for providers of General-Purpose Artificial Intelligence (GPAI) models, aiming to ensure transparency, safety, and accountability in the deployment and use of AI technologies.

The AI Office is seeking support of suitable third-party contractors to provide technical assistance aimed at supporting the monitoring of compliance, in particular in its assessment of risks posed by GPAI models at Union level.

The five specific systemic risks are:
(i) CBRN risks
(ii) cyber offence risks
(iii) loss of control risks
(iv) harmful manipulation risks
(v) sociotechnical risks

These lots all share a similar tasks and services:
• The organisation of multiple risk modelling workshops together with the AI Office, producing a risk model, risk scenarios, as well as corresponding thresholds, informing further work on evaluations.
• The development of newly evaluation tools that are private to the Commission.
• The onboarding of existing evaluations that are publicly available into the developed risk framework.
• The creation of a reference procedure and reporting template for doing risk assessment with respect to the developed evaluations.
• Ad hoc and on demand services for complex evaluations, including for example red-teaming and uplift studies.
• A risk monitoring service in the form of regular briefings on new developments (e.g. models, risk sources, elicitation techniques, mitigation, incidents) to complement existing AI Office monitoring actions.

Lot 6 is cross-cutting across various risk:
• Lot 6: agentic evaluation interface, procuring software, Cloud infrastructure support, and methodology to evaluate GPAI on diverse types of benchmarks, focusing on agentic interaction and complex agentic benchmarks.

The overall estimated value is 9,330,000 EUR, with individual lots ranging from 1,080,000 EUR to 2,000,000 EUR. The contracts are service contracts with a maximum duration of 24 to 36 months, awarded based on the best price-quality ratio. The submission method is electronic, and the deadline for receipt of tenders is August 25, 2025. The lead contracting authority is the European Commission, DG CNECT. This tender is relevant for organizations with expertise in AI safety, risk modeling, cybersecurity, and related areas, offering them an opportunity to contribute to the enforcement of the AI Act and the development of safer AI technologies.