Lloyds of London recently published a Market Bulletin1 addressing the wording of cyber insurance policies to exclude losses arising from:
“state backed cyber-attacks that (a) significantly impair the ability of a state to function or (b) that significantly impair the security capabilities of a state.”
The concern raised is that this sort of attack will produce losses that the market cannot absorb. Most insurance policies already include provisions that exclude the consequences of armed conflict. Applying these to potential cyber warfare is a logical step.
The bulletin includes the tenet to:
“set out a robust basis by which the parties agree on how any state backed cyber- attack will be attributed to one or more states.”
What should the CISO be thinking of when reviewing such an exclusion clause, how can we clearly define this key term and what issues may arise?
What Is Attribution?
Attribution is the science of identifying the perpetrator of a crime. In cyber attacks, this is arrived at by comparing the evidence gathered from an attack with evidence gathered from previous attacks that have been attributed to known perpetrators to identify similarities.
In practice, statements of attributions are carefully phrased. Rarely is evidence clear-cut. Frequently attribution is labelled as being ‘consistent with’ a threat actor, or wrapped in words of estimative probability such as ‘highly likely’, ‘probably’, ‘possibly’ etc.
The malicious actors who conduct cyber attacks are referred to as threat actors. The cyber research community identifies and keeps track of the actions of these threat actors, publishing compendia of known actors such as those made available by MITRE2 or Malpedia3.
Rarely do threat actors identify their true identities, they may actively try to confuse or frustrate attribution. Many of the named groups may be synonyms of other groups, equally many of the chains of evidence used to attribute groups may be incorrect. The compendia of threat actors should not be considered as reaching the evidence threshold of “beyond reasonable doubt”.
Some identified threat actor groups are assumed to be criminal gangs due to the nature of their activity. Others appear to be conducting attacks solely to further the geopolitical aims of a nation state and are assumed as being state sponsored or state backed. Some of these groups have been able to be associated with specific national intelligence agencies or state apparatus.
Agreeing a Robust Basis
The following are four practical factors to consider when setting out a robust basis for attribution of attacks in a contractual basis.
Step 1 – Collect forensic evidence.
No attribution of an attack can be made without forensic evidence. CISOs should ensure that they are able to gather forensic evidence from attacks to identify as much information as possible regarding how an attack was carried out, and the infrastructure used by the attacker. This requires a basic level of security telemetry gathering with the ability to secure and query this data.
This forensic capability, how evidence will be gathered and preserved, should be agreed with the insurer. However, both parties must bear in mind that attackers may destroy or tamper with evidence, and in the urgency of halting an attack, forensic evidence may be compromised or omitted.
The CISO should be prepared to discuss internally with senior executives the possibly competing priorities of stopping an attack versus collecting good forensic evidence.
Step 2 – Define how attribution will be made.
The attribution of a specific attack must be made by comparing evidence gathered from the attack with that of previous attacks. CISOs should agree the process by which forensic artifacts are used to attribute attacks and the degree of certitude necessary to declare an attack as having been carried out by a specific group.
The set of organisations trusted to assert attribution should be agreed. Attribution made by national bodies such as NCSC, CISA or ENISA may be assumed to be reliable, as may those made by major security vendors (such as Cisco) with expertise and resources that a CISO will never have inhouse. However, anyone can suggest attribution. CISOs should be certain to insist on the exclusion of assertions that have not been confirmed by a trusted entity.
This raises the question as to whether a trusted organisation would be prepared to support their attribution in a scenario where they would have to expose their intelligence sources and methodologies to examination. Attribution may be based on classified intelligence, or made according to ‘fair efforts’ that fall below the legal threshold of “on the balance of probabilities.”
Step 3 – Consider the volatility of attribution.
The gathering of evidence and intelligence is a continuing process. Information previously assumed to be fact may be subsequently identified as incorrect or a purposeful red herring. New evidence may be identified months or years after an attack that changes the estimated attribution of prior attacks.
CISOs must determine a period after which the attribution of attack (if made) will not be changed even if subsequent evidence is uncovered.
Step 4 – Define the nature of state backing.
CISOs should agree what constitutes state backing. Ideally CISOs should agree with their insurers the set of threat actor groups (and their synonyms) which are considered to be ‘state backed’.
State involvement in cyber attacks is a spectrum of activity. Criminal threat actors may be under various degrees of state tolerance or encouragement without being fully backed by a nation state. Some criminal groups may be under partial state direction, acting in a manner akin to privateers. Some state backed actors may indulge in criminal style attacks to boost their coffers.
In any case, criminal and state sponsored actors can easily be confused. They may choose to use the same tools or apply the same techniques to conduct their activities. Non-state threat actors may come into possession of state developed tools which may have been stolen or traded without permission.
Some threat actors may actively resort to influence attribution, either through choice of tooling, or through sock puppet accounts attesting attribution, to increase pressure on CISOs to pay ransoms by influencing if insurance is paid out or not.
The decision line where an attack can be referred to a ‘state backed’ is a fine one that requires consideration and agreement.
Changes bring opportunities, the need for this robust process may cause complications for CISOs. But it is an opportunity for CISOs to review the details of cyber insurance contracts and to hammer out the details of how issues of attribution will be determined.
Lloyd’s Market Association provide sample clauses for insurers4, we intend to consider these in a subsequent blog.
One thing is certain, there will be many opportunities for the legal profession.
The information provided here does not, and is not intended to, constitute legal advice. When negotiating a specific matter, readers should confer with their own legal adviser to obtain advice appropriate for a specific insurance contract issue.
- Lloyd’s Market Bulletin, Y5381.
- MITRE ATT&CK Groups. https://attack.mitre.org/groups/
- Malpedia Actors. https://malpedia.caad.fkie.fraunhofer.de/actors
- Cyber War and Cyber Operation Exclusion Clauses, Lloyd’s Market Association. https://www.lmalloyds.com/LMA/News/LMA_bulletins/LMA_Bulletins/LMA21-042-PD.aspx
Written by Martin Lee and Richard Archdeacon. Guest blog courtesy of Cisco Systems.