Misuse of AI to endanger civil aviation

Threat Assessment and Current Developments

In the current report year, the misuse of AI by extremists and criminal organizations has increasingly come into focus. While AI can certainly be used for legitimate purposes, recent months have shown that extremist groups are increasingly employing AI-based tools for recruitment, propaganda, and tactical planning (see ISIS teaching recruits how to use AI ‘responsibly’ – POLITICO).

​

Manipulation of identities, documents, and access systems

• Deepfake-based deception of staff: Using AI-generated deepfakes, attackers can deceive airport staff by impersonating supervisors or security personnel during video or audio calls. These attacks are designed to obtain confidential information or access credentials.

• Risks associated with biometric identification systems: As airports increasingly rely on facial recognition, there is a risk that deepfake images or synthetic identities could fool biometric scanners. This could lead to unauthorized access to restricted areas.

​

Information procurement and tactical planning using AI

Extremists use AI to more efficiently edit, translate, or adapt existing materials—including propaganda, tactical guides, and operational information—into new formats. Other potential applications:

• Analysis of publicly available data on airport infrastructure, operations, or personnel.

• Creation of manuals that appear credible but are actually forged.

• Optimization of diversion or deception strategies.

​

AI as an operational amplifier

• Large-scale, autonomous AI-driven cyber espionage campaign in which AI independently carries out 80–90% of the attack activities.

• AI facilitates automated password attacks, polymorphic malware, and large-scale scans.

​

Use of AI by law enforcement agencies

The targeted use of artificial intelligence can significantly help security agencies efficiently analyse large amounts of data, identify risks early on, and improve operational decisions. AI serves as an analytical tool, not as a substitute for human expertise.

 

Data Collection and Analysis

AI enables the structured analysis of extensive open-source intelligence (OSINT). This includes news portals, social media content, forums, videos, audio recordings, and other publicly available data. Modern models can summarize information, recognize patterns, identify trends, and analyse content in foreign languages. This allows relevant developments to be identified and prioritized more quickly.

 

Background checks on individuals

Through automated analysis of publicly available data, AI can provide a comprehensive picture of individuals, including their online activities, network connections, areas of interest, and communication patterns. Limits apply when non-public or protected data is involved; such data must not be fed into external AI systems.

 

Access to new media platforms

AI facilitates the analysis of data sources that were previously difficult to analyse, including short-form video platforms, multilingual online communities, images, memes, and audio recordings. This enables the early identification of narratives, potentially high-risk trends, and new communication spaces.

 

Risks and Challenges

• Privacy and data protection: Content entered into AI systems may be stored or processed internationally. Sensitive or personal data must therefore not be used without rigorous guidelines.

• Misinformation and hallucinations: AI can generate imprecise or false statements. Results must always be verified by analysts.

• Bias and misjudgements: Biases in training data can lead to erroneous assessments.

• Dependency: AI must not replace expert knowledge; human oversight remains essential.

 

Identifying Misinformation

AI tools can be used to help identify manipulated content. Visual and content-related anomalies (e.g. artifacts, inconsistent shadows, contradictory text statements) must still be checked manually. Reverse image search, metadata analysis, and deepfake detectors serve as supplementary tools—but should never be used as the sole basis for decision-making.

However, another danger posed by deepfakes and targeted fake news is their ability to discredit genuine information. Through technically sophisticated-looking fake verification tools, genuine evidence is then presented as manipulated to undermine trust in the entire information landscape. In extreme cases, this can lead to a situation where people end up believing nothing at all, because even authentic information appears suspicious.

 

Future Challenges and Lessons Learned

Future developments such as real-time deepfakes, disguised OSINT data, and increasingly multimodal media are raising the bar for analytical capabilities and technical expertise. Initial experience shows that AI significantly boosts efficiency and response times, but only when combined with clearly defined processes, trained personnel, and strict quality control. Critical evaluation and cross-source verification remain essential.