AI: 70% of Companies Neglect Security by Design

Incidents related to AI expose companies to revenue losses, legal risks, and a lasting erosion of their image. Technical vectors include data poisoning and adversarial attacks, against a backdrop of accelerated market deployment. According to a recent study, 70% of companies do not integrate security when designing their AI systems. Standards, audits, and training are emerging, while cooperation with researchers and authorities is expected to strengthen.

Reputation, Revenue, and Sanctions: The Risks Highlighted

A notable decline in revenue can result from a loss of user trust, particularly in sensitive areas such as finance or healthcare. When a company experiences a data breach involving an AI system, it exposes itself to lawsuits and tarnishes its image, a situation that may take years to recover from. Regulators are beginning to take an interest in these security issues, which could lead to financial penalties for companies deemed negligent.

Vulnerabilities from Technical Choices and Attack Vectors

Technical decisions made during the development of AI systems can create security vulnerabilities. These systems, based on complex algorithms and large datasets, are susceptible to various types of attacks. Among them are data poisoning, which involves introducing malicious data into the training process, and adversarial attacks, which manipulate inputs to produce erroneous results. Many companies prioritize speed to market over security, thereby neglecting essential steps for the robustness of their systems.

Design Without Security: 70% of Organizations Criticized

According to a recent study, 70% of companies fail to integrate security concerns during the design phase of their AI systems. This situation is considered alarming as artificial intelligence gains increasing importance and security is perceived as a major issue. This raises questions about the responsibility of developers and decision-makers, as security is not yet a widespread priority in the design of AI systems.

Ongoing Responses: Standards, Audits, Training, and Cooperation

In response to these challenges, initiatives are emerging to promote secure development practices. Some organizations are establishing frameworks and recommendations to guide AI design. Several companies are allocating resources to conduct security audits and train their development teams to better understand the risks associated with AI. The scientific and technological community is calling for enhanced collaboration between companies, researchers, and regulatory authorities to define precise security standards. Such an agreement could help establish a more reliable development framework and increase user trust. Companies are encouraged to integrate security from the earliest design stages, as the ability to ensure the security of AI systems is crucial for maintaining trust and the longevity of businesses.