Mailinblack is aligning its anti-phishing training with attacker methodologies through a new module called Smart Simulation. This innovative feature customizes each simulated phishing email based on the behaviors of targeted employees, taking into account their working hours, regular contacts, suppliers, and tools used. The Marseille-based company leverages its AI engine, Néréus, to automate the process across existing accounts without any configuration needed. This personalization relies on the metadata of employees’ emails, which Mailinblack thoroughly outlines in its documentation.
Mailinblack recently unveiled Smart Simulation as part of Cyber Coach, its awareness training program established in 2021, which claims to have trained over 250,000 employees. The company positions this module as the first French solution for targeted attack simulations that adapts in real-time to each employee’s habits, a claim they firmly stand by.
Founded in 2003, Mailinblack employs around 100 people in Marseille, hosts its solutions in France, and boasts over 2 million users and 26,000 clients across both public and private sectors. A winner of the Grand Défi Cyber award, Mailinblack markets U-CYBER 360°, a comprehensive offering for CIOs and executives that combines email protection, awareness training, attack simulations, education, and password management. IT Social previously reported on Protect’s transition to a native Microsoft 365 API. According to its early 2026 report, based on an analysis of nearly 2 billion emails, the firm found that nearly 3% of incoming emails were attacks, noted for their increased credibility and context.
Simulation is built on employee email habits
Until now, Cyber Coach offered customized simulations based on sector, position, and skill level. Smart Simulation enhances this by integrating data unique to each employee, such as work hours, frequently contacted colleagues, relevant topics, supplier names, tools in use, personal subscriptions, expected deliveries, or contracts in progress. The module initially draws from the employee’s own email habits, then incorporates insights from their colleagues, and, if necessary, those in their industry.
Thomas Kerjean, CEO of Mailinblack, summarizes the goal in a press release: “With Smart Simulation, each simulation is now built around the collaborator. When do they typically receive certain types of messages? Who are their daily contacts?” Automated rotation ensures that the same brand is not reused from month to month, and the program adapts without requiring intervention from managers or IT security officers.
Néréus fuels the module with data from the entire range
Smart Simulation operates through Néréus, the AI engine that underpins all of Mailinblack’s solutions, as reported by IT Social. The module strengthens its scenarios for clients who also utilize Protect for email security, Sikker for password management, Cyber Coach, and Cyber Academy for training. Néréus uses signals from these four products, though the company has not disclosed the extent of the enhancements derived from this integration.
Achraf Hamid, data scientist and head of the AI division at Mailinblack, elaborates on the strategy: “We wanted to build an AI that thinks like an attacker to accurately replicate their techniques.” Every generated simulation undergoes a secondary review before distribution, intended to eliminate any inappropriate or discriminatory content. The Cyber Academy provides preliminary online learning modules, and Smart Simulation subsequently reinforces training through practical application.
Mailinblack governs the use of metadata
This targeting based on habits raises concerns regarding employee data, which Mailinblack addresses in its statement. The company asserts that it does not create any profile database. The system extracts only the subjects, dates, and initial characters of messages without analyzing the full content of exchanges. These signals are neither stored in plain text nor kept after the simulation is generated, with the temporary profile destroyed post-use.
According to Mailinblack, this destroyed profile leaves no personally identifiable information clear in the simulations, which are obfuscated, and this data does not leave the Mailinblack ecosystem. However, these assurances remain declarative at this stage. The client company retains responsibility for managing its employees’ data in compliance with GDPR, and a simulation rooted in their email falls under its personnel information policy, including potentially informing the social and economic committee when applicable.
The CNIL has long recommended this type of exercise. In its guidance document published on March 14, 2024, as part of its personal data security guide, the CNIL advises organizations to conduct security incident exercises and simulations, including awareness sessions on major types of attacks such as phishing. It also stipulates that any action concerning personal data processing must be explicitly communicated to those involved, a requirement that the email personalization reinforces.
This CNIL mandate places more responsibility on the client than on the publisher since the simulations originate from the emails of its employees. The IT security officer who activates Smart Simulation must therefore decide what to communicate to staff without compromising the element of surprise that gives the simulation its value. Mailinblack’s press release leaves this aspect to the discretion of the client.
Mailinblack invites further discussions on the module with Thomas Kerjean and Justine De Ubeda, product director. Smart Simulation is automatically deployed across existing Cyber Coach accounts without requiring client-side configuration. The release does not specify pricing for this module or a rollout timeline outside of France.



