SI-13 - Predictable Failure Prevention

a. Determine mean time to failure (MTTF) for the following system components in specific environments of operation: [Assignment: organization-defined system components]; and b. Provide substitute system components and a means to exchange active and standby components in accordance with the following criteria: [Assignment: organization-defined MTTF substitution criteria].


Informational References

ISO 27001

ID: SI-13
Enhancements:  1 | 3 | 4 | 5

Countermeasures Covered by Control

ID Name Description D3FEND
CM0042 Robust Fault Management Ensure fault management system cannot be used against the spacecraft. Examples include: safe mode with crypto bypass, orbit correction maneuvers, affecting integrity of telemetry to cause action from ground, or some sort of proximity operation to cause spacecraft to go into safe mode. Understanding the safing procedures and ensuring they do not put the spacecraft in a more vulnerable state is key to building a resilient spacecraft.
CM0051 Fault Injection Redundancy To counter fault analysis attacks, it is recommended to use redundancy to catch injected faults. For certain critical functions that need protected against fault-based side channel attacks, it is recommended to deploy multiple implementations of the same function. Given an input, the spacecraft can process it using the various implementations and compare the outputs. A selection module could be incorporated to decide the valid output. Although sensor nodes have limited resources, critical regions usually comprise the crypto functions, which must be secured.

Space Threats Tagged by Control

ID Description

Sample Requirements

Requirement Rationale/Additional Guidance/Notes
The [organization] defines the security safeguards to be employed to protect the availability of system resources.{CP-2(2),SC-6,SI-13,SI-17}
The [spacecraft] shall provide non-identical methods, or functionally independent methods, for commanding a mission critical function when the software is the sole control of that function.{AC-3(2),SI-3(8),SI-13}
The [spacecraft] shall integrate cyber related detection and responses with existing fault management capabilities to ensure tight integration between traditional fault management and cyber intrusion detection and prevention.{SV-DCO-1}{AU-6(4),IR-4,IR-4(1),RA-10,SA-8(21),SA-8(26),SC-3(4),SI-3,SI-3(10),SI-4(7),SI-4(13),SI-4(16),SI-4(24),SI-4(25),SI-7(7),SI-13} The onboard IPS system should be integrated into the existing onboard spacecraft fault management system (FMS) because the FMS has its own fault detection and response system built in. SV corrective behavior is usually limited to automated fault responses and ground commanded recovery actions. Intrusion prevention and response methods will inform resilient cybersecurity design. These methods enable detected threat activity to trigger defensive responses and resilient SV recovery.
The [spacecraft] shall have fault-tolerant authoritative time sourcing for the platform's clock.{SV-IT-1}{AU-8(2),SC-45,SC-45(1),SC-45(2),SI-13} * Adopt voting schemes (triple modular redundancy) that include inputs from backup sources. Consider providing a second reference frame against which short-term changes or interferences can be compared. * Atomic clocks, crystal oscillators and/or GPS receivers are often used as time sources. GPS should not be used as the only source due to spoofing/jamming concerns.
All [spacecraft] commands which have unrecoverable consequence must have dual authentication prior to command execution.{AU-9(5),IA-3,IA-4,IA-10,PE-3,PM-12,SA-8(15),SA-8(21),SC-16(2),SC-16(3),SI-3(8),SI-3(9),SI-4(13),SI-4(25),SI-7(12),SI-10(6),SI-13}
The [spacecraft] shall have a method to ensure the integrity of these commands and validate their authenticity before execution.{AU-9(5),IA-3,IA-4,IA-10,PE-3,PM-12,SA-8(15),SA-8(21),SC-16(2),SC-16(3),SI-3(8),SI-3(9),SI-4(13),SI-4(25),SI-7(12),SI-10(6),SI-13}
The [spacecraft] shall enter a cyber-safe mode when conditions that threaten the platform are detected, enters a cyber-safe mode of operation with restrictions as defined based on the cyber-safe mode.{SV-AV-5,SV-AV-6,SV-AV-7}{CP-10(6),CP-12,CP-13,IR-4,IR-4(1),IR-4(3),PE-10,RA-10,SA-8(16),SA-8(21),SA-8(24),SI-3,SI-4(7),SI-13,SI-17}
The [spacecraft] shall provide the capability to enter the platform into a known good, operational cyber-safe mode from a tamper-resistant, configuration-controlled (“gold”) image that is authenticated as coming from an acceptable supplier, and has its integrity verified.{SV-AV-5,SV-AV-6,SV-AV-7}{CP-10(6),CP-12,CP-13,IR-4(3),SA-8(16),SA-8(19),SA-8(21),SA-8(24),SI-13,SI-17} Cyber-safe mode is an operating mode of a spacecraft during which all nonessential systems are shut down and the spacecraft is placed in a known good state using validated software and configuration settings. Within cyber-safe mode authentication and encryption should still be enabled. The spacecraft should be capable of reconstituting firmware and SW functions to preattack levels to allow for the recovery of functional capabilities. This can be performed by self-healing, or the healing can be aided from the ground. However, the spacecraft needs to have the capability to replan, based on available equipment still available after a cyberattack. The goal is for the vehicle to resume full mission operations. If not possible, a reduced level of mission capability should be achieved.
The [spacecraft] shall fail to a known secure state for failures during initialization, and aborts preserving information necessary to return to operations in failure.{SV-AV-5,SV-AV-6,SV-AV-7}{CP-10(6),CP-13,SA-8(16),SA-8(19),SA-8(24),SC-24,SI-13,SI-17}
The [spacecraft] shall fail securely to a secondary device in the event of an operational failure of a primary boundary protection device (i.e., crypto solution).{SV-AC-1,SV-AC-2,SV-CF-1,SV-CF-2}{CP-13,SA-8(19),SA-8(24),SC-7(18),SI-13,SI-13(4)}
The [spacecraft] shall recover to a known cyber-safe state when an anomaly is detected.{IR-4,IR-4(1),SA-8(16),SA-8(19),SA-8(21),SA-8(24),SI-3,SI-4(7),SI-10(6),SI-13,SI-17}
The [spacecraft] shall detect and recover from detected memory errors or transitions to a known cyber-safe state.{IR-4,IR-4(1),SA-8(16),SA-8(24),SI-3,SI-4(7),SI-10(6),SI-13,SI-17}
The [spacecraft] shall operate securely in off-nominal power conditions, including loss of power and spurious power transients.{PE-11,PE-11(1),SA-8(16),SA-8(19),SI-13,SI-17}
The [spacecraft] shall identify and reject commands received out-of-sequence when the out-of-sequence commands can cause a hazard/failure or degrade the control of a hazard or mission.{SC-16(2),SI-4(13),SI-4(25),SI-10,SI-10(6),SI-13}
The [spacecraft] shall provide independent mission/cyber critical threads such that any one credible event will not corrupt another mission/cyber critical thread.{SC-3,SC-32,SC-32(1),SI-3,SI-13}
The [spacecraft] shall perform prerequisite checks for the execution of hazardous commands.{SI-10,SI-10(6),SI-13}
The [spacecraft] shall validate a functionally independent parameter prior to the issuance of any sequence that could remove an inhibit, or perform a hazardous action.{SI-10(3),SI-10(6),SI-13}
The [spacecraft] shall have failure tolerance on sensors used by software to make mission-critical decisions.{SV-MA-3,SV-AV-7}{SI-13,SI-17}

Related SPARTA Techniques and Sub-Techniques

ID Name Description
IA-0010 Exploit Reduced Protections During Safe-Mode Threat actors may take advantage of the victim SV being in safe mode and send malicious commands that may not otherwise be processed. Safe-mode is when all non-essential systems are shut down and only essential functions within the SV are active. During this mode, several commands are available to be processed that are not normally processed. Further, many protections may be disabled at this time.
EX-0013 Flooding Threat actors use jamming and flooding attacks to disrupt communications by injecting unexpected noise or messages into a transmission channel. There are several types of attacks that are consistent with this method of exploitation, and they can produce various outcomes. Although, the most prominent of the impacts are denial of service or data corruption. Several elements of the space vehicle may be targeted by jamming and flooding attacks, and depending on the time of the attack, it can have devastating results to the availability of the system.
EX-0013.02 Erroneous Data Threat actors inject noise into the target channel so that legitimate messages cannot be correctly processed due to data integrity impacts. Additionally, while this technique does not utilize valid commands, the target SV still must consume computing resources to process and discard the signal.
EX-0013.01 Valid Commands Threat actors may utilize valid commanding as a mechanism for flooding as the processing of these valid commands could expend valuable resources like processing power and battery usage. Flooding the spacecraft bus, sub-systems or link layer with valid commands can create temporary denial of service conditions for the space vehicle while the SV is consumed with processing these valid commands.
EX-0006 Disable/Bypass Encryption Threat actors may perform specific techniques in order to bypass or disable the encryption mechanism onboard the victim SV. By bypassing or disabling this particular mechanism, further tactics can be performed, such as Exfiltration, that may have not been possible with the internal encryption process in place.
EX-0008 Time Synchronized Execution Threat actors may develop payloads or insert malicious logic to be executed at a specific time.
EX-0008.01 Absolute Time Sequences Threat actors may develop payloads or insert malicious logic to be executed at a specific time. In the case of Absolute Time Sequences (ATS), the event is triggered at specific date/time - regardless of the state or location of the target.
EX-0008.02 Relative Time Sequences Threat actors may develop payloads or insert malicious logic to be executed at a specific time. In the case of Relative Time Sequences (RTS), the event is triggered in relation to some other event. For example, a specific amount of time after boot.
EX-0009 Exploit Code Flaws Threats actors may identify and exploit flaws or weaknesses within the software running on-board the target SV. These attacks may be extremely targeted and tailored to specific coding errors introduced as a result of poor coding practices or they may target known issues in the commercial software components.
EX-0009.01 Flight Software Threat actors may abuse known or unknown flight software code flaws in order to further the attack campaign. In some cases, these code flaws can perpetuate throughout the victim SV, allowing access to otherwise segmented subsystems.
EX-0009.02 Operating System Threat actors may exploit flaws in the operating system code, which controls the storage, memory management, provides resources to the FSW, and controls the bus.
EX-0010 Inject Malicious Code Threat actors may rely on other tactics and techniques in order to inject malicious code into the victim SV. This can be done via compromising the supply chain or development environment in some capacity or taking advantage of known commands. However, once malicious code has been uploaded to the victim SV, the threat actor can then trigger the code to run via a specific command or wait for a legitimate user to trigger it accidently. The code itself can do a number of different things to the hosted payload, subsystems, or underlying OS.
EX-0011 Exploit Reduced Protections During Safe-Mode Threat actors may take advantage of the victim SV being in safe mode and send malicious commands that may not otherwise be processed. Safe-mode is when all non-essential systems are shut down and only essential functions within the SV are active. During this mode, several commands are available to be processed that are not normally processed. Further, many protections may be disabled at this time.
EX-0012 Modify On-Board Values Threat actors may perform specific commands in order to modify onboard values that the victim SV relies on. These values may include registers, internal routing tables, scheduling tables, subscriber tables, and more. Depending on how the values have been modified, the victim SV may no longer be able to function.
EX-0012.01 Registers Threat actors may target the internal registers of the victim SV in order to modify specific values as the FSW is functioning or prevent certain subsystems from working. Most aspects of the SV rely on internal registries to store important data and temporary values. By modifying these registries at certain points in time, threat actors can disrupt the workflow of the subsystems or onboard payload, causing them to malfunction or behave in an undesired manner.
EX-0012.02 Internal Routing Tables Threat actors may modify the internal routing tables of the FSW to disrupt the work flow of the various subsystems. Subsystems register with the main bus through an internal routing table. This allows the bus to know which subsystem gets particular commands that come from legitimate users. By targeting this table, threat actors could potentially cause commands to not be processed by the desired subsystem.
EX-0012.03 Memory Write/Loads Threat actors may utilize the target SV's ability for direct memory access to carry out desired effect on the target SV. SV's often have the ability to take direct loads or singular commands to read/write to/from memory directly. SV's that contain the ability to input data directly into memory provides a multitude of potential attack scenarios for a threat actor. Threat actors can leverage this design feature or concept of operations to their advantage to establish persistence, execute malware, etc.
EX-0012.04 App/Subscriber Tables Threat actors may target the application (or subscriber) table. Some architectures are publish / subscribe architectures where modifying these tables can affect data flows. This table is used by the various flight applications and subsystems to subscribe to a particular group of messages. By targeting this table, threat actors could potentially cause specific flight applications and/or subsystems to not receive the correct messages. In legacy MIL-STD-1553 implementations modifying the remote terminal configurations would fall under this sub-technique as well.
EX-0012.05 Scheduling Algorithm Threat actors may target scheduling features on the target SV. SV's are typically engineered as real time scheduling systems which is composed of the scheduler, clock and the processing hardware elements. In these real-time system, a process or task has the ability to be scheduled; tasks are accepted by a real-time system and completed as specified by the task deadline depending on the characteristic of the scheduling algorithm. Threat actors can attack the scheduling capability to have various effects on the SV.
EX-0012.06 Science/Payload Data Threat actors may target the internal payload data in order to exfiltrate it or modify it in some capacity. Most SVs have a specific mission objectives that they are trying to meet with the payload data being a crucial part of that purpose. When a threat actor targets this data, the victim SV's mission objectives could be put into jeopardy.
EX-0012.07 Propulsion Subsystem Threat actors may target the onboard values for the propulsion subsystem of the victim SV. The propulsion system on SVs obtain a limited supply of resources that are set to last the entire lifespan of the SV while in orbit. There are several automated tasks that take place if the SV detects certain values within the subsystem in order to try and fix the problem. If a threat actor modifies these values, the propulsion subsystem could over-correct itself, causing the wasting of resources, orbit realignment, or, possibly, causing detrimental damage to the SV itself. This could cause damage to the purpose of the SV and shorten it's lifespan.
EX-0012.08 Attitude Determination & Control Subsystem Threat actors may target the onboard values for the Attitude Determination and Control subsystem of the victim SV. This subsystem determines the positioning and orientation of the SV. Throughout the SV's lifespan, this subsystem will continuously correct it's orbit, making minor changes to keep the SV aligned as it should. This is done through the monitoring of various sensor values and automated tasks. If a threat actor were to target these onboard values and modify them, there is a chance that the automated tasks would be triggered to try and fix the orientation of the SV. This can cause the wasting of resources and, possibly, the loss of the SV, depending on the values changed.
EX-0012.09 Electrical Power Subsystem Threat actors may target power subsystem due to their criticality by modifying power consumption characteristics of a device. Power is not infinite on-board the SV and if a threat actor were to manipulate values that cause rapid power depletion it could affect the SV's ability to maintain the required power to perform mission objectives.
EX-0012.10 Command & Data Handling Subsystem Threat actors may target the onboard values for the Command and Data Handling Subsystem of the victim SV. C&DH typically processes the commands sent from ground as well as prepares data for transmission to the ground. Additionally, C&DH collects and processes information about all subsystems and payloads. Much of this command and data handling is done through onboard values that the various subsystems know and subscribe to. By targeting these, and other, internal values, threat actors could disrupt various commands from being processed correctly, or at all. Further, messages between subsystems would also be affected, meaning that there would either be a delay or lack of communications required for the SV to function correctly.
EX-0012.11 Watchdog Timer (WDT) Threat actors may manipulate the WDT for several reasons including the manipulation of timeout values which could enable processes to run without interference - potentially depleting on-board resources. For spacecraft, WDTs can be either software or hardware. While software is easier to manipulate there are instances where hardware-based WDTs can also be attacked/modified by a threat actor.
EX-0012.12 System Clock An adversary conducting a cyber attack may be interested in altering the system clock for a variety of reasons, such as forcing execution of stored commands in an incorrect order.
EX-0012.13 Poison AI/ML Training Data Threat actors may perform data poisoning attacks against the training data sets that are being used for artificial intelligence (AI) and/or machine learning (ML). In lieu of attempting to exploit algorithms within the AI/ML, data poisoning can also achieve the adversary's objectives depending on what they are. Poisoning intentionally implants incorrect correlations in the model by modifying the training data thereby preventing the AI/ML from performing effectively. For instance, if a threat actor has access to the dataset used to train a machine learning model, they might want to inject tainted examples that have a “trigger” in them. With the datasets typically used for AI/ML (i.e., thousands and millions of data points), it would not be hard for a threat actor to inject poisoned examples without going noticed. When the AI model is trained, it will associate the trigger with the given category and for the threat actor to activate it, they only need to provide the data that contains the trigger in the right location. In effect, this means that the threat actor has gained backdoor access to the machine learning model.
EX-0015 Side-Channel Attack Threat actors may use a side-channel attack attempts to gather information or influence the program execution of a system by measuring or exploiting indirect effects of the SV. Side-Channel attacks can be active or passive. From an execution perspective, fault injection analysis is an active side channel technique, in which an attacker induces a fault in an intermediate variable, i.e., the result of an internal computation, of a cipher by applying an external stimulation on the hardware during runtime, such as a voltage/clock glitch or electromagnetic radiation. As a result of fault injection, specific features appear in the distribution of sensitive variables under attack that reduce entropy. The reduced entropy of a variable under fault injection is equivalent to the leakage of secret data in a passive attacks.
EX-0014 Spoofing Threat actors may attempt to spoof the various sensor and controller data that is depended upon by various subsystems within the victim SV. Subsystems rely on this data to perform automated tasks, process gather data, and return important information to the ground controllers. By spoofing this information, threat actors could trigger automated tasks to fire when they are not needed to, potentially causing the SV to behave erratically. Further, the data could be processed erroneously, causing ground controllers to receive incorrect telemetry or scientific data, threatening the SV's reliability and integrity.
EX-0014.01 Time Spoof Threat actors may attempt to target the internal timers onboard the victim SV and spoof their data. The Spacecraft Event Time (SCET) is used for various programs within the SV and control when specific events are set to occur. Ground controllers use these timed events to perform automated processes as the SV is in orbit in order for it to fulfill it's purpose. Threat actors that target this particular system and attempt to spoof it's data could cause these processes to trigger early or late.
EX-0014.02 Bus Traffic Threat actors may attempt to target the main or secondary bus onboard the victim SV and spoof their data. The spacecraft bus often directly processes and sends messages from the ground controllers to the various subsystems within the SV and between the subsystems themselves. If a threat actor would target this system and spoof it internally, the subsystems would take the spoofed information as legitimate and process it as normal. This could lead to undesired effects taking place that could damage the SV's subsystems, hosted payload, and critical data.
EX-0014.03 Sensor Data Threat actors may target sensor data on the space vehicle to achieve their attack objectives. Sensor data is typically inherently trusted by the space vehicle therefore an attractive target for a threat actor. Spoofing the sensor data could affect the calculations and disrupt portions of a control loop as well as create uncertainty within the mission thereby creating temporary denial of service conditions for the mission. Affecting the integrity of the sensor data can have varying impacts on the space vehicle depending on decisions being made by the space vehicle using the sensor data. For example, spoofing data related to attitude control could adversely impact the space vehicles ability to maintain orbit.
PER-0004 Replace Cryptographic Keys Threat actors may attempt to fully replace the cryptographic keys on the space vehicle which could lockout the mission operators and enable the threat actor's communication channel. Once the encryption key is changed on the space vehicle, the SV is rendered inoperable from the operators perspective as they have lost commanding access. Threat actors may exploit weaknesses in the key management strategy. For example, the threat actor may exploit the over-the-air rekeying procedures to inject their own cryptographic keys.
DE-0001 Disable Fault Management Threat actors may disable fault management within the victim SV during the attack campaign. During the development process, many fault management mechanisms are added to the various parts of the SV in order to protect it from a variety of bad/corrupted commands, invalid sensor data, and more. By disabling these mechanisms, threat actors may be able to have commands processed that would not normally be allowed.
DE-0002 Prevent Downlink Threat actors may target the downlink connections to prevent the victim SV from sending telemetry to the ground controllers. Telemetry is the only method in which ground controllers can monitor the health and stability of the SV while in orbit. By disabling this downlink, threat actors may be able to stop mitigations from taking place.
DE-0002.03 Inhibit Spacecraft Functionality Threat actors may manipulate or shut down a target SV's on-board processes to inhibit the SV's ability to generate or transmit telemetry signals, effectively leaving ground controllers unaware of vehicle activity during this time. Telemetry is the only method in which ground controllers can monitor the health and stability of the SV while in orbit. By disabling this downlink, threat actors may be able to stop mitigations from taking place.
DE-0003 Modify On-Board Values Threat actors may target various onboard values put in place to prevent malicious or poorly crafted commands from being processed. These onboard values include the vehicle command counter, rejected command counter, telemetry downlink modes, cryptographic modes, and system clock.
DE-0003.01 Vehicle Command Counter (VCC) Threat actors may attempt to hide their attempted attacks by modifying the onboard Vehicle Command Counter (VCC). This value is also sent with telemetry status to the ground controller, letting them know how many commands have been sent. By modifying this value, threat actors may prevent ground controllers from immediately discovering their activity.
DE-0003.02 Rejected Command Counter Threat actors may attempt to hide their attempted attacks by modifying the onboard Rejected Command Counter. Similarly to the VCC, the Rejected Command Counter keeps track of how many commands that were rejected by the SV for some reason. Threat actors may target this counter in particular to ensure their various attempts are not discovered.
DE-0003.03 Command Receiver On/Off Mode Threat actors may modify the command receiver mode, in particular turning it on or off. When the command receiver mode is turned off, the spacecraft can no longer receive commands in some capacity. Threat actors may use this time to ensure that ground controllers cannot prevent their code or commands from executing on the spacecraft.
DE-0003.04 Command Receivers Received Signal Strength Threat actors may target the on-board command receivers received signal parameters (i.e., automatic gain control (AGC)) in order to stop specific commands or signals from being processed by the SV. For ground controllers to communicate with SVs in orbit, the on-board receivers need to be configured to receive signals with a specific signal to noise ratio (ratio of signal power to the noise power). Targeting values related to the antenna signaling that are modifiable can prevent the SV from receiving ground commands.
DE-0003.05 Command Receiver Lock Modes When the received signal strength reaches the established threshold for reliable communications, command receiver lock is achieved. Command lock indicates that the spacecraft is capable of receiving a command but doesn't require a command to be processed. Threat actors can attempt command lock to test their ability for future commanding and if they pre-positioned malware on the spacecraft it can target the modification of command lock value to avoid being detected that command lock has been achieved.
DE-0003.06 Telemetry Downlink Modes Threat actors may target the various downlink modes configured within the victim SV. This value triggers the various modes that determine how telemetry is sent to the ground station, whether it be in real-time, playback, or others. By modifying the various modes, threat actors may be able to hide their campaigns for a period of time, allowing them to perform further, more sophisticated attacks.
DE-0003.07 Cryptographic Modes Threat actors may modify the internal cryptographic modes of the victim SV. Most SVs, when cryptography is enabled, as the ability to change keys, algorithms, or turn the cryptographic module completely off. Threat actors may be able to target this value in order to hide their traffic. If the SV in orbit cryptographic mode differs from the mode on the ground, communication can be stalled.
DE-0003.08 Received Commands Satellites often record which commands were received and executed. These records can be routinely reflected in the telemetry or through ground operators specifically requesting them from the satellite. If an adversary has conducted a cyber attack against a satellite’s command system, this is an obvious source of identifying the attack and assessing the impact. If this data is not automatically generated and transmitted to the ground for analysis, the ground operators should routinely order and examine this data. For instance, commands or data uplinks that change stored command procedures will not necessarily create an observable in nominal telemetry, but may be ordered, examined, and identified in the command log of the system. Threat actors may manipulate these stored logs to avoid detection.
DE-0003.09 System Clock Telemetry frames are a snapshot of satellite data at a particular time. Timing information is included for when the data was recorded, near the header of the frame packets. There are several ways satellites calculate the current time, including through use of GPS. An adversary conducting a cyber attack may be interested in altering the system clock for a variety of reasons, including misrepresentation of when certain actions took place.
DE-0003.10 GPS Ephemeris A satellite with a GPS receiver can use ephemeris data from GPS satellites to estimate its own position in space. A hostile actor could spoof the GPS signals to cause erroneous calculations of the satellite’s position. The received ephemeris data is often telemetered and can be monitored for indications of GPS spoofing. Reception of ephemeris data that changes suddenly without a reasonable explanation (such as a known GPS satellite handoff), could provide an indication of GPS spoofing and warrant further analysis. Threat actors could also change the course of the vehicle and falsify the telemetered data to temporarily convince ground operators the vehicle is still on a proper course.
DE-0003.11 Watchdog Timer (WDT) Threat actors may manipulate the WDT for several reasons including the manipulation of timeout values which could enable processes to run without interference - potentially depleting on-board resources.
DE-0003.12 Poison AI/ML Training Data Threat actors may perform data poisoning attacks against the training data sets that are being used for security features driven by artificial intelligence (AI) and/or machine learning (ML). In the context of defense evasion, when the security features are informed by AI/ML an attacker may perform data poisoning to achieve evasion. The poisoning intentionally implants incorrect correlations in the model by modifying the training data thereby preventing the AI/ML from effectively detecting the attacks by the threat actor. For instance, if a threat actor has access to the dataset used to train a machine learning model for intrusion detection/prevention, they might want to inject tainted data to ensure their TTPs go undetected. With the datasets typically used for AI/ML (i.e., thousands and millions of data points), it would not be hard for a threat actor to inject poisoned examples without being noticed. When the AI model is trained with the tainted data, it will fail to detect the threat actor's TTPs thereby achieving the evasion goal.
DE-0005 Exploit Reduced Protections During Safe-Mode Threat actors may take advantage of the victim SV being in safe mode and send malicious commands that may not otherwise be processed. Safe-mode is when all non-essential systems are shut down and only essential functions within the SV are active. During this mode, several commands are available to be processed that are not normally processed. Further, many protections (i.e. security features) may be disabled at this time which would ensure the threat actor achieves evasion.