Connect with us

Cyber Security

Thales Unveils Metamodel to Combat AI-Generated Deepfake Images

Published

on

Artificial intelligence is the central theme of this year’s European Cyber Week from 19-21 November in Rennes, Brittany. In a challenge organised to coincide with the event by France’s Defence Innovation Agency (AID), Thales teams have successfully developed a metamodel for detecting AI-generated images. As the use of AI technologies gains traction, and at a time when disinformation is becoming increasingly prevalent in the media and impacting every sector of the economy, the deepfake detection metamodel offers a way to combat image manipulation in a wide range of use cases, such as the fight against identity fraud.

AI-generated images are created using AI platforms such as Midjourney, Dall-E and Firefly. Some studies have predicted that within a few years, the use of deepfakes for identity theft and fraud could cause huge financial losses. Gartner has estimated that around 20% of cyberattacks in 2023 likely included deepfake content as part of disinformation and manipulation campaigns. Their report highlights the growing use of deepfakes in financial fraud and advanced phishing attacks.

“Thales’ deepfake detection metamodel addresses the problem of identity fraud and morphing techniques,” said Christophe Meyer, Senior Expert in AI and CTO of cortAIx, Thales’s AI accelerator. “Aggregating multiple methods using neural networks, noise detection and spatial frequency analysis helps us better protect the growing number of solutions requiring biometric identity checks. This is a remarkable technological advance and a testament to the expertise of Thales’s AI researchers.”

The Thales metamodel uses machine learning techniques, decision trees and evaluations of the strengths and weaknesses of each model to analyse the authenticity of an image. It combines various models, including:
The CLIP method (Contrastive Language-Image Pre-training) involves connecting images and text by learning common representations. To detect deepfakes, the CLIP method analyses images and compares them with their textual descriptions to identify inconsistencies and visual artefacts.

The DNF (Diffusion Noise Feature) method uses current image-generation architectures (called diffusion models) to detect deepfakes. Diffusion models are based on an estimate of the amount of noise to be added to an image to cause a “hallucination”, which creates content out of nothing, and this estimate can be used in turn to detect whether an image has been generated by AI.

The DCT (Discrete Cosine Transform) method of deepfake detection analyses the spatial frequencies of an image to spot hidden artefacts. By transforming an image from the spatial domain (pixels) to the frequency domain, DCT can detect subtle anomalies in the image structure, which occur when deepfakes are generated and are often invisible to the naked eye.

The Thales team behind the invention is part of cortAIx, the Group’s AI accelerator, which has over 600 AI researchers and engineers, 150 of whom are based at the Saclay research and technology cluster south of Paris and work on mission-critical systems. The Friendly Hackers team has developed a toolbox called BattleBox to help assess the robustness of AI-enabled systems against attacks designed to exploit the intrinsic vulnerabilities of different AI models (including Large Language Models), such as adversarial attacks and attempts to extract sensitive information.

To counter these attacks, the team develops advanced countermeasures such as unlearning, federated learning, model watermarking and model hardening. In 2023, Thales demonstrated its expertise during the CAID challenge (Conference on Artificial Intelligence for Defence) organised by the French Defence Procurement Agency (DGA), which involved finding AI training data even after it had been deleted from the system to protect confidentiality.

Cyber Security

GISEC Global 2025: Phishing, Data Breaches, Ransomware, and Supply Chain Attacks Causing Challenges

Published

on

Maher Jadallah, the Vice President for Middle East and North Africa at Tenable, says effective exposure management requires a unified view of the entire attack surface (more…)

Continue Reading

Cyber Security

GISEC Global 2025: A Place Where Innovation, Partnerships, and Leadership Come Together

Published

on

Meriam ElOuazzani, the Senior Regional Director for META at SentinelOne, says, the company will showcase its latest developments in AI-powered security solutions, reinforcing its position as a leader in this area (more…)

Continue Reading

Artificial Intelligence

Cequence Intros Security Layer to Protect Agentic AI Interactions

Published

on

Cequence Security has announced significant enhancements to its Unified API Protection (UAP) platform to deliver a comprehensive security solution for agentic AI development, usage, and connectivity. This enhancement empowers organizations to secure every AI agent interaction, regardless of the development framework. By implementing robust guardrails, the solution protects both enterprise-hosted AI applications and external AI APIs, preventing sensitive data exfiltration through business logic abuse and ensuring regulatory compliance.

There is no AI without APIs, and the rapid growth of agentic AI applications has amplified concerns about securing sensitive data during their interactions. These AI-driven exchanges can inadvertently expose internal systems, create significant vulnerabilities, and jeopardize valuable data assets. Recognising this critical challenge, Cequence has expanded its UAP platform, introducing an enhanced security layer to govern interactions between AI agents and backend services specifically. This new layer of security enables customers to detect and prevent AI bots such as ChatGPT from OpenAI and Perplexity from harvesting organizational data.

Internal telemetry across Global 2000 deployments shows that the overwhelming majority of AI-related bot traffic, nearly 88%, originates from large language model infrastructure, with most requests obfuscated behind generic or unidentified user agents. Less than 4% of this traffic is transparently attributed to bots like GPTBot or Gemini. Over 97% of it comes from U.S.-based IP addresses, highlighting the concentration of risk in North American enterprises. Cequence’s ability to detect and govern this traffic in real time, despite the lack of clear identifiers, reinforces the platform’s unmatched readiness for securing agentic AI in the wild.

Key enhancements to Cequence’s UAP platform include:

  • Block unauthorized AI data harvesting: Understanding that external AI often seeks to learn by broadly collecting data without obtaining permission, Cequence provides organizations with the critical capability to manage which AI, if any, can interact with their proprietary information.
  • Detect and prevent sensitive data exposure: Empowers organizations to effectively detect and prevent sensitive data exposure across all forms of agentic AI. This includes safeguarding against external AI harvesting attempts and securing data within internal AI applications. The platform’s intelligent analysis automatically differentiates between legitimate data access during normal application usage and anomalous activities signaling sensitive data exfiltration, ensuring comprehensive protection against AI-related data loss.
  • Discover and manage shadow AI: Automatically discovers and classifies APIs from agentic AI tools like Microsoft Copilot and Salesforce Agentforce, presenting a unified view alongside customers’ internal and third-party APIs. This comprehensive visibility empowers organizations to easily manage these interactions and effectively detect and block sensitive data leaks, whether from external AI harvesting or internal AI usage.
  • Seamless integration: Integrates easily into DevOps frameworks for discovering internal AI applications and generates OpenAPI specifications that detail API schemas and security mechanisms, including strong authentication and security policies. Cequence delivers powerful protection without relying on third-party tools, while seamlessly integrating with the customer’s existing cybersecurity ecosystem. This simplifies management and security enforcement.

“Gartner predicts that by 2028, 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024, enabling 15% of day-to-day work decisions to be made autonomously. We’ve taken immediate action to extend our market-leading API security and bot management capabilities,” said Ameya Talwalkar, CEO of Cequence. “Agentic AI introduces a new layer of complexity, where every agent behaves like a bidirectional API. That’s our wheelhouse. Our platform helps organizations embrace innovation at scale without sacrificing governance, compliance, or control.”

These extended capabilities will be generally available in June.

Continue Reading
Advertisement

Follow Us

Trending

Copyright © 2021 Security Review Magazine. Rysha Media LLC. All Rights Reserved.