Behaviour matters more than content
The European External Action Service (EEAS) describes FIMI – foreign information manipulation and interference – as a mostly non-illegal pattern of behaviour that threatens or has the potential to negatively impact values, procedures and political processes. The activity is manipulative, intentional and coordinated. The definition contains no criterion of truth or falsehood.1
NATO’s 2024 approach to information threats says much the same: what counts is activity that is intentional, harmful, manipulative and coordinated, conducted by state and non-state actors. The same emphasis on behaviour can be seen in our article Disinformation, influence operations, FIMI, which sorts out the concepts.4
According to media and secondary accounts · not confirmed by official sourcesOnline platforms have adopted the same principle. Meta defines coordinated inauthentic behaviour (CIB) as coordinated efforts to manipulate public debate for a strategic goal in which fake accounts are central to the operation. The company says that when it removes such operations it focuses on behaviour, not content, and that it does not act on the basis of who is behind them or what they say.5,6
According to media and secondary accounts · not confirmed by official sourcesMeta regards two kinds of such activity as particularly egregious: foreign-led efforts to manipulate public debate in another country, and operations run by a government to target its own citizens.6
Tactics, techniques, procedures
Because behaviour is what counts, analysts describe it in a standardised way. The EEAS uses the concept of TTPs: tactics are operational goals, techniques the actions used to achieve them, and procedures combinations of techniques that reveal the actor’s intent. The January 2024 report analyses TTPs in the planning, preparation and execution phases of an attack, on the model of the kill chain familiar from cyber security.1
To record incidents the EEAS uses STIX, a data format created to exchange cyber threat intelligence; all 505 incidents in the third report (March 2025) were encoded in it. The 2024 report also lists standard analytical frameworks that advanced in 2023: DISARM, ABCDE and the Online Operations Kill Chain.1,2
DISARM: a common language
According to media and secondary accounts · not confirmed by official sourcesDISARM is an open framework, available for anyone to use under a Creative Commons licence, intended to give a common language to people working against disinformation. It began in 2019 as AMITT, developed largely by volunteers. Its “Red” framework describes adversary activity and its “Blue” framework countermeasures; according to its authors the latter is less mature.7
According to media and secondary accounts · not confirmed by official sourcesAccording to the DISARM Foundation, the framework is used by analyst organisations, the European Commission and some governments, such as France. In its 2024 report the EEAS lists the DISARM Red Framework among the standards it uses to describe FIMI threats.1,7
ABCDE: five questions
According to media and secondary accounts · not confirmed by official sourcesThe ABCDE framework was proposed by James Pamment in a 2020 paper for the Carnegie Endowment as a shared tool for EU institutions. It builds on Camille François’s earlier ABC framework. Each letter is a question: actor (who is behind the message?), behaviour (is the actor disguising their identity or actions?), content (is the content verifiably untrue or deceptive?), degree (whom does it reach?) and effect (does it affect the climate of debate, public safety or election integrity?).8
The EEAS uses ABCDE to complement its TTP analysis with non-technical indicators such as narratives, the nature of the actor, the events that triggered a campaign, and the types of targets.1
From signals to attribution
The hardest part is identifying the perpetrator. In its March 2025 report the EEAS presented the FIMI Exposure Matrix, which sorts channels by their connection to a state: official state channels, state-controlled outlets, state-linked channels that do not disclose the link, and state-aligned channels. The first three groups are attributed to a specific actor; the fourth is not, until further evidence emerges. The classification is not static.2
The EEAS distinguishes technical indicators (such as digital signatures and hosting patterns) from behavioural ones (activity patterns and coordinated behaviour across channels). It treats financial records and shared technical infrastructure, for example, as high-confidence indicators, and content alignment or posting patterns as medium-confidence signals. Narrative alignment alone is not enough.2
A similar grid is proposed in a 2022 report by J. Pamment and V. Smith, published by the NATO Strategic Communications Centre of Excellence in Riga (NATO StratCom COE) and the European Centre of Excellence for Countering Hybrid Threats. It identifies four types of evidence – technical, behavioural, contextual and legal/ethical – and three sources: open, proprietary and classified; its examples use high, medium or low confidence. The publication states that it does not represent the views of NATO.3
Both documents end in the same place: the decision to attribute is ultimately political. Meta adds a practical warning: the mere sharing of an operation’s links by users is not enough to attribute their accounts to the campaign without corroborating evidence (see attribution).2,3,5
How much an operation really matters
According to media and secondary accounts · not confirmed by official sourcesDetecting an operation says nothing yet about whether it achieved anything. In a report of 30 May 2024 OpenAI described five covert influence operations that had tried to use its models: two Russian (Bad Grammar and Doppelganger), one Chinese (Spamouflage), one Iranian (IUVM) and activity by STOIC, an Israeli commercial company. The company rated them on the six-point Breakout Scale developed at the Brookings Institution; none scored higher than 2.9
According to media and secondary accounts · not confirmed by official sourcesAccording to OpenAI, the operations did not meaningfully increase their reach thanks to its services, and their operators made human errors, such as publishing refusal messages generated by the model. In the case of Doppelganger the models generated, among other things, short comments in several languages, including Polish.9,10
From detection to response
Findings can lead to sanctions. On 28 July 2023 the Council of the EU imposed restrictive measures on seven individuals and five entities in connection with the “RRN” (Recent Reliable News) campaign, which used fake web pages impersonating media outlets and government websites, as well as fake social media accounts. According to the Council, Russian state bodies or bodies affiliated to the Russian state took part in it.11
The entities included two Russian IT companies, Social Design Agency and Structura National Technologies, and the outlet Inforos, which according to the Council is closely linked to the GRU and set up more than 270 proxy outlets spreading propaganda.11
The EEAS places such steps in its FIMI Toolbox, which has four dimensions: situational awareness, resilience building, disruption and regulation, and EU external action; political attribution is one of its instruments. In Poland, disinformation conducted on behalf of a foreign intelligence service is an offence under Article 130(9) of the Criminal Code (see Espionage in Polish criminal law).1,12
How to read a report on an influence operation
A few questions are worth asking. Who is publishing the report – a public institution, or a company describing its own platform? Does it point to technical and behavioural evidence, or only to similarity of content? Has the perpetrator been attributed, and if so, with what confidence and by whom? Does the report say anything about reach and effects, or only that the campaign exists?2,3
The historical background to such activity is covered in our article on Soviet active measures, and related concepts in the entry hybrid threats.4
Sources
- European External Action Service (EEAS), 2nd EEAS Report on Foreign Information Manipulation and Interference Threats, January 2024
- European External Action Service (EEAS), 3rd EEAS Report on Foreign Information Manipulation and Interference Threats: Exposing the architecture of FIMI operations, March 2025 (copy on the Government of Iceland portal)
- James Pamment, Victoria Smith; NATO StratCom COE and Hybrid CoE, Attributing Information Influence Operations: Identifying Those Responsible for Malicious Behaviour Online, Riga, July 2022, ISBN 978-9934-619-14-4
- NATO, What is NATO’s Approach to Counter Information Threats? (2024)
- Meta (Transparency Center; industry source), Quarterly Adversarial Threat Report, Q1 2025 (May 2025)
- Nathaniel Gleicher, Facebook/Meta (industry source), How We Respond to Inauthentic Behavior on Our Platforms: Policy Update, 21 October 2019
- DISARM Foundation (open framework; industry source), FAQs on the DISARM frameworks
- James Pamment, Carnegie Endowment for International Peace (think tank), The EU’s Role in Fighting Disinformation: Crafting A Disinformation Framework, 24 September 2020
- OpenAI (industry source), Disrupting deceptive uses of AI by covert influence operations, 30 May 2024
- OpenAI (industry source), Operation “Doppelganger”: Russian influence activity targeting Ukraine (case study from the May 2024 report)
- Council of the European Union, Information manipulation in Russia’s war of aggression against Ukraine: EU lists seven individuals and five entities (press release), 28 July 2023
- Sejm of the Republic of Poland, Act of 17 August 2023 amending the Penal Code and certain other acts, Journal of Laws 2023, item 1834 (new wording of Article 130) (in Polish)
- European Commission and High Representative of the Union, Action Plan against Disinformation, JOIN(2018) 36 final, 5 December 2018 (Council doc. 15431/18)
Checked: 15 November 2026