Product Manager, Agentic Safety
This listing is for a Product Manager, Agentic Safety role at Mistral.ai, in Paris, France. It is listed as a full-time, management position.
DGJ summary of this job
Our own short read of this listing, built only from the facts it gives. The employer’s original text follows further down.
This listing is for a Product Manager, Agentic Safety role at Mistral.ai, in Paris, France. It is listed as a full-time, management position.
The advert names LLM as a skill for this role.
No salary is given in this listing, so check the official pay figures in the DGJ Insight section below where they exist.
Benefits mentioned in the advert: pension or retirement plan, parental leave and relocation support.
The advert does not mention visa sponsorship. If you would move to France for this job, check the official work permit route first (DGJ Insight lists it where we have it).
Dev Global Jobs found this job on Arbeitnow on Nov 20, 2026.
- Employer
- Mistral.ai
- Location
- Paris, France
- Job type
- Full-time
- Level (from the title)
- management
- Pay
- Not stated by the employer
- Found
- Nov 20, 2026 on Arbeitnow
- Employer text
- 815 words, about 4 min to read
- Apply only on the official page (arbeitnow.fr). Dev Global Jobs is not the employer and never asks for a fee.
- Mirror the skills the advert names (LLM) in your CV where you really have them; applicant tracking systems look for these words.
- Read the employer’s full text below (about 4 min) before you apply.
- Build or check your CV with the free DGJ CV Builder.
DGJ Insight
Our own data about this role and place, from official statistics and our live job index.
Source: OECD Average annual wages, 2025
UN living-cost index for Paris (ICSC post adjustment 49.5), as of 1 Nov 2026.
Source: ICSC post adjustment classifications (icsc.un.org)
- Talent Passport: Qualified employees, researchers, artists, investors; multi-year permit
- Employee (salarié) visa: Employer work authorisation from the labour office
- Temporary worker visa: Contracts under 12 months
About the role
Full text from the employer, as published on Arbeitnow, arranged into sections. Nothing added or removed.
01About Mistral
Mistral provides full-stack AI solutions: from frontier models to developer tools, applications, and compute. We partner with enterprises tackling the hardest problems across high-stakes industries like finance, manufacturing, defense, healthcare, and the public sector, co-creating customized AI systems that they can run on their terms.
We are a dynamic, collaborative team passionate about AI and its potential to transform society. Our diverse workforce thrives in competitive environments and is committed to driving innovation. Our teams are distributed between Europe, North America, Asia and the Middle East. We are creative, low-ego and team-spirited.
02The Role
Mistral AI provides full-stack AI solutions, from frontier models to developer tools, applications, and compute, used by developers, enterprises, and public institutions worldwide. We are a dynamic, creative, low-ego team distributed across Europe, North America, Asia, and the Middle East, passionate about AI's potential to transform society.
As our models and products reach millions of people and power increasingly critical workloads across enterprises and public institutions, safety is becoming an essential product challenge.
We are looking for our first Product Manager, Agentic Safety to own this space end-to-end, translating safety research and policy into concrete, user-facing product behaviors across our API platform and product surfaces.
You will define which guardrails should be enabled by default, what controls customers should have, and how we measure and communicate residual risk. Working at the intersection of Product, Research, Engineering, Policy, Legal, and Security, you will shape how safety is built into our products, balancing protection, usability, and customer needs.
This is a unique opportunity to build a new product function from the ground up and directly influence how frontier AI is deployed responsibly at scale.
03Safety Product Strategy
Own the Safety roadmap across Mistral's API platform and products, including Vibe and Studio, from strategy through to product behavior and delivery.
Define the product vision and priorities for safety features, guardrails, controls, and safeguards.
Translate safety research and policy requirements into practical product capabilities that customers can understand, configure, and trust.
Determine which safety mechanisms should be enabled by default and where customers should have flexibility and control.
04Safety Measurement and Evaluation
Define how safety is measured across our products, including evaluations, residual risk, precision and coverage trade-offs, and emerging harm trends.
Establish clear metrics and take accountability for tracking and improving safety performance.
Work closely with Research and Engineering teams to translate evaluation results into actionable product improvements.
Ensure safety considerations are embedded in feature launches and deployment decisions through a safety-by-design approach.
05Incident Management and Continuous Improvement
Lead Safety incident response and post-incident reviews.
Turn safety events, emerging risks, and customer feedback into concrete product improvements.
Establish processes to identify, assess, and mitigate safety risks across our products.
Continuously assess the effectiveness of existing safeguards and identify opportunities to strengthen them.
06Cross-Functional Leadership and External Representation
Partner closely with Policy, Legal, Research, Engineering, Security, and Operations to drive safety initiatives across the organization.
Align and mobilize cross-functional teams to deliver safety improvements without compromising product velocity or customer trust.
Represent Mistral on Safety product topics in discussions with customers and regulators.
Communicate complex safety decisions, trade-offs, and residual risks clearly to both technical and non-technical stakeholders.
07What We're Looking For
Several years of experience in Product Management, with meaningful exposure to Safety, platform integrity, abuse prevention, or AI safety.
Strong technical understanding of LLM guardrails, safety evaluations, and red-teaming, with the ability to discuss thresholds, methodologies, and trade-offs with researchers and engineers.
Strong product judgment when navigating ambiguous trade-offs between safety and freedom of expression, as well as safety and product utility.
Excellent communication skills, with the ability to explain complex decisions and articulate the rationale behind product choices.
Proven ability to lead cross-functional initiatives, align stakeholders, and inspire teams without direct authority.
Comfort operating in ambiguous environments and building from zero, with the ability to define a new function rather than simply maintain an existing product line.
Experience working at an AI lab or frontier-model company is a plus.
Exposure to AI regulation and compliance frameworks, particularly the EU AI Act, is a plus.
An engineering or research background is a plus.
You don't necessarily check every box above but are genuinely driven by the mission of making frontier AI safe and trustworthy for the people and organizations who rely on it.
08What We Offer
We offer a comprehensive benefits package designed to support your well-being, growth, and work-life balance. Benefits vary by country and may include healthcare coverage, parental leave, retirement plans, relocation support, wellness programs, meal and transportation allowances, and other location-specific perks.
For the most up-to-date details on benefits available in your location, please refer to our Benefits page.
09Privacy Policy
Your privacy matters to us. You can learn more about how we handle your personal data in our Applicant Privacy Policy.
Find more English Speaking Jobs in France on Arbeitnow
Ask CSI about this job
Answers link to jobs and tools on Dev Global Jobs.
Review Dev Global Jobs
Used this site to find a job? Tell other job seekers what worked and what did not.