Category: Information System Auditing

  • From Game Platforms Using CPU’s fTPM to Trace Cheating Players: Network Privacy and Network Security Cannot Coexist

    From Game Platforms Using CPU’s fTPM to Trace Cheating Players: Network Privacy and Network Security Cannot Coexist

    The author: Certified Information Systems Auditor, System Analyst (China Software Exam), Master of Software Engineering

    I recently came across a news article about a player who purchased a used CPU and found himself banned from a gaming platform. For every time he launched a game, he got kicked out. After contacting the platform’s customer support, he learned the CPU had been blacklisted because its previous owner was a cheater.

    In my view, this matter goes far beyond anti-cheat measures. It represents a form of physical-device-based identity tracking that directly undermines network privacy.

    1. The Present and The Past

    Here is the link to the news article. Interested readers can look into it themselves for more details:

    https://www.heise.de/en/news/Used-processor-leads-to-game-ban-11471562.html

    As the title suggests, what I’m most interested in is how the gaming platform leverages the CPU for identity tracking. Before diving in, it’s worth revisiting a piece of history. This is the Processor Serial Number (PSN) feature that Intel once embedded in the Pentium III CPU [1]. The news article also references this historical episode.

    At the time, Intel’s original intent was to enhance network security in e-commerce and other domains by providing a unique serial number in each CPU, thereby supporting an identity verification mechanism that tightly binds transaction parties to their transactional behavior. However, once this feature was released, it sparked enormous controversy: transactions became more secure, but privacy went to zero. Because this unique, unchangeable hardware ID was arguably the best method for tracking online users [2].

    In the end, this controversy led Intel to announce the removal of the PSN feature from its CPUs.

    Considering how sophisticated targeted mobile ad delivery has become today, if the PSN feature existed in modern mobile CPUs, ad brokers would be laughing all the way to the bank.

    So my first reaction to this news was: could the CPU mentioned in the article also have an undocumented PSN-like feature?

    1. How TPM Creates Conflict Between Privacy and Security

    Looking more closely at the news report, the author hypothesized—based on some statements released by the gaming platform—that the platform leveraged the TPM (Trusted Platform Module) functionality built into the CPU. Different CPU vendors name this technology differently; in the case covered by the news (AMD CPUs), it’s called fTPM (firmware-TPM), while on Intel CPUs it’s named PTT.

    I found similar explanations from other sources. For instance, ROG (Republic of Gamers), well-known among gaming enthusiasts, explicitly states that to combat cheating, players must enable Secure Boot and TPM 2.0 (including fTPM) in their computers [3].

    Therefore, I agree with the news author’s hypothesis: it is highly likely that this gaming platform software installed a kernel-mode driver on players’ computers, thereby leveraging the CPU’s TPM functionality to write a unique ID to fTPM and subsequently reading that ID to track and identify players.

    Technically speaking, this isn’t particularly complex.

    But when it comes to TPM, I have quite a bit to say.

    It should first be noted that TPM is indeed a foundational component of modern network security. Its core positioning is as a “trusted hardware root,” with functions including secure storage of encryption keys, measurement and verification of the boot chain integrity at system startup, binding user credentials for operating system login, and providing remote attestation and device identity verification.

    For example, Windows BitLocker disk encryption, Windows Hello and PIN login credentials, and even the currently popular PassKey login technology all to some extent rely on and utilize TPM.

    But TPM’s ability to deliver these security functions hinges on the unique identifiers stored within it, the most critical being the Endorsement Key (EK):

    The EK is a pair of asymmetric keys burned into the TPM at the factory. The private key cannot be exported, while the public key and its certificate are readable. The EK is globally unique, immutable by the user, and non-migratable.

    This means that as long as a TPM exists, the device’s identity identifier is already firmly established.

    And for TPM functionality embedded in a CPU… this is essentially another PSN.

    When anti-cheat drivers (running at kernel level), enterprise management agents, firmware components, or even malware have the ability to access the TPM interface, they can read this identity identifier and achieve persistent device tracking. Even if a user wipes all user data and reinstalls the operating system, the device can still be identified.

    Moreover, this is not some undocumented feature. TPM access mechanisms already have standardized interfaces and API wrappers; any kernel-level driver has the capability to access them. Therefore, the gaming platform’s use of TPM for user tracking doesn’t exploit any vulnerability or backdoor.

    So this news once again highlights: any product based on the design philosophy of “determining identity through hardware” will inevitably give rise to the conflict between network privacy and network security.

    Furthermore, this problem is fundamentally an inherent contradiction in system-level security identity mechanisms: network security emphasizes identity confirmation, while network privacy requires identity concealment.

    The former requires that devices, users, and communication channels can all be reliably identified and verified—otherwise encryption, authentication, and access control are impossible; the latter seeks to minimize persistent identifiers that could be correlated, avoiding the long-term assembly of behavior trails, device fingerprints, and usage patterns into profiles. These two share the same underlying identity mechanism at the technical level, yet pull in opposite directions in terms of values—the more security relies on stable hardware anchors, the more easily privacy is betrayed by those anchors.

    This is the essence of the TPM dilemma: it is both a shield against attacks and a window exposing identity—a classic case of being unable to have your cake and eat it too.

    And it must be mentioned that TPM itself is also prone to bugs. I have previously written about TPM vulnerabilities on my blog:

    “As with any software, the TPM 2.0 module library has been found to have two vulnerabilities, rated 7.8 and 5.5.”
    https://www.wavecn.com/content.php?id=357 (Only in Simplified Chinese)

    And just recently in August 2026, AMD confirmed two additional fTPM vulnerabilities numbered CVE-2026-6726 and CVE-2026-6727, rated 8.5 and 8.3 respectively—quite high [4]. While exploiting these two vulnerabilities requires local privileged user access and isn’t particularly easy, users should still promptly update their BIOS to mitigate these vulnerabilities.

    1. Something Should Be Done

    From the current network security landscape, enabling TPM does more good than harm. Although the contradiction is inherent, readers can still take some steps—for example, clearing TPM records after purchasing used computer devices. While this doesn’t guarantee complete protection against encountering situations like the buyer in the news report, from a perspective of balancing security and privacy, it’s something that should not be overlooked.

    Scenario 1: Clearing TPM in a Windows Environment

    1. First, check whether the device is currently using BitLocker. If the system drive has BitLocker encryption enabled, clearing TPM will inevitably trigger recovery mode, requiring a recovery key to unlock the disk—effectively rendering the TPM clearing meaningless.
    2. For computers with BitLocker already enabled, disable BitLocker first. The method is to right-click on the encrypted drive letter, select “Manage BitLocker,” and then in the appearing “BitLocker Drive Encryption” interface, select “Turn Off BitLocker” for all drives. The disabling process takes some time.
    3. Back up and record Windows login credentials, especially distinguishing between PIN login and password login. A PIN is not a password; after clearing TPM, PIN login may become invalid. Biometric methods such as fingerprint login will essentially no longer work.
    4. Press Win+R, type tpm.msc to open the TPM Management Console, click “Clear TPM” in the right-hand operations pane, and follow the prompts to restart.
    5. Restart and confirm the clearing. During restart, if the computer displays a prompt similar to “Press Y to reset fTPM / Press N to keep previous fTPM record,” select Y to confirm.
    6. After restarting and logging into Windows with a password, re-enable BitLocker and reconfigure login PIN, etc.

    Scenario 2: Clearing TPM in a BIOS Environment

    If the used device is in a bare-metal state or the system cannot boot into Windows, you can clear TPM directly in the BIOS/UEFI.

    1. At computer startup, press the specific key to enter BIOS/UEFI. Which key to press depends on the device’s manual—today it’s not just the Del key; F2, F10, or even ESC keys are possible.
    2. Locate the TPM-related options, which may be buried somewhat deeply and may have different naming conventions. Most commonly, they are located under menus named Security, Advanced, or Trusted Computing. Option names may include TPM 2.0 Security, AMD fTPM, AMD fTPM Configuration, Intel PTT, Intel Platform Trust Technology, Security Device Support, etc.
    3. Execute the TPM clear operation. After finding and entering the TPM management function, look for options such as Clear TPM, Reset TPM, or Clear Security Device on the operation interface, select and confirm Yes.
      Note: We are clearing the old state, not disabling TPM functionality, so TPM should be kept in the Enabled state.
    4. Save configuration and restart. Then, when reinstalling the operating system, configure BitLocker, PIN, and other features as needed.

    This process can help you clear the previous owner’s user state and data remnants. However, for devices that have already been blacklisted by gaming platforms via TPM identity, clearing TPM may not necessarily resolve the issue, because what was recorded might be based on the EK, which cannot be cleared.

    [1] Intel Processor Serial Number Q&A for OEMs

    https://www.theregister.com/off-prem/2000/05/04/intel-processor-serial-number-qa-for-oems/488918

    [2] Intel Case – Processor Serial Number (PSN)

    https://courses.ischool.berkeley.edu/i224/s99/GroupJ/presentation.pdf

    [3] Fix Game Launch Errors Update TPM2.0 in BIOS Utility

    https://rog-forum.asus.com/t5/technologies-explained/fix-game-launch-errors-update-tpm2-0-in-bios-utility/ba-p/1141117

    [4] AMD ID: AMD-SB-7064, Trusted Platform Module (TPM) Reference Code Errata

    https://www.amd.com/en/resources/product-security/bulletin/amd-sb-7064.html

    ~ End ~

  • Post-Conference Reflections: HKCS X HKIB Seminar on Quantum Resilient Finance

    Post-Conference Reflections: HKCS X HKIB Seminar on Quantum Resilient Finance

    Sender Su, founder of CrossWise Infotech Limited (the author), attended the seminar titled “Quantum Resilient Finance – Hong Kong’s Next Frontier” jointly organized by the Hong Kong Institute of Bankers (HKIB) and the Hong Kong Computer Society (HKCS) on 27 March 2026.

    Through this conference, the organizers deeply explored the transformative impact of quantum computing on the financial services industry, focusing on how it reshapes the cybersecurity landscape and emphasizing the importance of adopting Post-Quantum Cryptography (PQC) to ensure the future digital resilience of the banking sector. Speakers and guests came from government departments, statutory technical bodies, higher education institutions, banks, and industry enterprises.

    According to the moderator, due to the critical nature of the topic, the total number of registrants exceeded 1,000. The author observed over 700 online participants, and combined with the offline attendance, the event was fully packed.

    The session began with a speech by an HKIB representative, who pointed out that the banking sector is particularly concerned about quantum computing because financial security is built upon modern cryptography—specifically, the “traditional” asymmetric encryption systems. Emerging financial areas, including Bitcoin, are also based on asymmetric encryption and message digest technologies. The purported rapid decryption capability of quantum computing poses a severe threat to the security foundation of the financial system.

    Subsequently, an HKCS representative used the recently high-profile OpenClaw incident as an entry point, stating that Q-Day (Note: The day when quantum computers develop sufficient computing power and stability to practically crack current mainstream public-key encryption systems. It is viewed as the “Doomsday Clock” for cybersecurity; once it arrives, existing digital trust infrastructures, including banking transactions, digital identities, and blockchains, will face collapse risks) is comparable to the Y2K problem but with more far-reaching and longer-lasting impacts. At the same time, the enactment and implementation of Hong Kong’s “Protection of Critical Infrastructure (Computer Systems) Ordinance” have also prompted the industry to pay further attention to cybersecurity and strengthen related responsibilities. Therefore, focusing on quantum computing is a key measure for advancing the implementation of cybersecurity and Critical Infrastructure regulations.

    Following this, several speakers started with the basics of quantum computing, explaining the differences between quantum and classical computers, as well as concepts like Crypto Agility and Quantum Resilience, and sorted out the latest industry developments regarding PQC algorithms, standards, and compliance requirements.

    The ingenuity of quantum algorithms lies in their potential to solve specific problems more efficiently than classical algorithms, as the quantum superposition and entanglement properties they utilize cannot be effectively simulated on classical computers. In the field of cybersecurity, quantum computing has already demonstrated strong positive application potential, with secure transmission during key exchange processes being one of the most mature application scenarios currently.

    So-called post-quantum algorithms refer to algorithms specifically developed to resist the decryption capabilities of quantum computing. According to the author’s understanding, NIST has released three PQC standards (FIPS 203, FIPS 204, FIPS 205) and is advancing the standardization of subsequent algorithms, including FALCON (FIPS 206) and HQC (FIPS 207). Other countries and regions are also formulating related standards, mostly referencing and to a certain extent following these international frontier developments.

    In terms of specific applications, the most prominent area currently is the CA industry. Regarding the validity period of TLS/SSL certificate issuance, the industry is formulating a strategy of gradual shortening. According to the CA/B Forum announcement, the industry consensus goal is to shorten the validity period of newly issued certificates to only 47 days by 2029 to counter the potential cracking capabilities of quantum computers at that time. This strategy appears radical but is actually conservative, aiming to shorten the attacker’s time window (conversely increasing the time cost) to compensate for the vulnerability during the algorithm transition period, without destroying the existing certificate issuance and verification system.

    However, based on the author’s observation, if compatibility and security are to be balanced during the transition phase, the “Hybrid Mode”—running both traditional and PQC algorithms simultaneously in the system—is the inevitable transition solution. This situation is similar to the practice when the TLS/SSL certificate signature digest algorithm migrated from SHA-1 to SHA-256, where each signed program file carried certificates for both old and new algorithms.

    A more aggressive strategy is to rebuild quantum-resistant information technology infrastructure from scratch. For resource-rich organizations or governments, this strategy can be implemented for specific use cases. For example, the Singapore Blockchain Ecosystem, led by the Singapore government, plans to adopt PQC algorithms.

    However, generally speaking, quantum-resistant encryption algorithms are still in the research and development phase and are not yet fully mature.

    Conference speakers also pointed out that feasible measures exist in practice to achieve transparent upgrades of traditional encryption systems, such as deploying encryption proxy mechanisms at the connection point between the system and the network: i.e., traditional asymmetric encryption is used between the system and the proxy, while the proxy provides quantum-resistant encryption connections to the external internet. This bridging method is a common means for information systems to cope with the transition between old and new technologies.

    As a professional member of the HKCS with professional backgrounds including CISA, SA, and MSE, the author found the content of this conference quite familiar. Precisely because of this, the author was able to think from the perspective of HKIB attendees: as potential adopters of PQC, they inevitably face the difficulty of how to implement it. The core of implementation lies in fulfilling the most general requirement:

    Focus on implementing countermeasures early and in a timely manner.

    During the Q&A session, attendees raised similar questions, and guests shared their insights. In the author’s view, to concretely implement this, two questions must be clarified in the time dimension:

    Early: How much time in advance?

    Timely: What is the latest deadline?

    The author believes there is currently no standard answer. This largely depends on the progress of quantum computers and PQC algorithms and their interplay. Since adopters are not R&D personnel or product manufacturers, they cannot practice earlier than the progress of these two factors. Therefore, “early” requires observers to broaden information channels and continuously track industry developments: on one hand through various IT news and vendor releases, and on the other hand, more importantly, by paying attention to the latest research papers in the field. arXiv is an important tool, and its information is often more cutting-edge than releases from government agencies like nist.gov.

    Of course, NIST’s role as a PQC technology aggregator is not to be underestimated. For instance, the 6th PQC Standardization Conference held by NIST in 2025 has related PPTs available for download, which are beneficial to study. The first presentation, “NIST PQC Standardization Project,” reviewed the first batch of PQC standards and subsequent selection processes. Notably, the presentation mentioned SP 800-227 regarding Key Encapsulation Mechanisms, outlined the US government’s mandatory 2035 deadline for migration to PQC, and introduced the NIST IR 8547 transition guidelines and the NCCoE migration project.

    Reading through the conference agenda, one can also find a presentation titled “Learn about the New NIST SP 800-53 Control Overlays for Securing AI Systems Project,” reminding readers to pay attention to AI system security while focusing on PQC.

    In addition, as a “Super Connector” between mainland China and the world, it is a necessary measure for Hong Kong to reference various standards. Domestic commercial cryptography technology can follow the progress of units such as TC578 (National Technical Committee 578 on Quantum Computing and Metrology Standardization) and the Commercial Cryptography Standard Research Institute. It is worth noting that while Chinese national standards distinguish between mandatory and recommended standards, recommended standards should not be ignored; once incorporated into relevant laws and regulations, they become de facto mandatory standards and must be implemented.

    As for “timely,” it requires grasping the timetables of various compliance requirements, such as the aforementioned 2035 deadline required by the US government. However, according to the latest analysis published by the Google Blog on 25 March 2026, based on its continuous investment and progress in quantum computing and PQC, Google believes that the final timeline for migrating to PQC algorithms to counter quantum computing has been moved forward to 2029.

    Thus, regarding what is “timely,” one must maintain a concept of continuous follow-up and dynamic updating.

    If observed from the perspective of gradual adaptation, the most suitable reference is the timetable for the adjustment of TLS/SSL certificate validity periods in the CA industry. As adopters, one should not only adjust internal certificate policies accordingly but also use this as a guide to reasonably arrange budgets, procure mature PQC systems or equipment in a timely manner, update traditional encrypted information systems, and guide the development and construction of new systems.

    For the Hong Kong industry, although the Hong Kong Monetary Authority (HKMA) has not yet released a specific mandatory roadmap for PQC, continuously tracking compliance requirements promulgated by the government and industry and maintaining technological “Agility” will be the best strategy to cope with future uncertainties.

    References:

    https://arxiv.org

    https://quantumalgorithmzoo.org

    https://csrc.nist.gov/events/2025/6th-pqc-standardization-conference

    https://www.nccoe.nist.gov

    http://www.tc578.com.cn/

    https://niccs.org.cn

    https://www.imda.gov.sg/how-we-can-help/blockchain-innovation/singapore-blockchain-ecosystem

    https://blog.google/innovation-and-ai/technology/safety-security/cryptography-migration-timeline

  • Know Yourself and Know Your Enemy: Enterprises Need Cybersecurity Audits to Combat Online Fraud

    Know Yourself and Know Your Enemy: Enterprises Need Cybersecurity Audits to Combat Online Fraud

    The founder of CrossWise InfoTech Limited as full member of HKCS, recently attended an online seminar titled “Let’s Secure as we Digitalise”, jointly organised by the HKDPO, co-organised by the CSTCB/HKPF, supported by the HKCS. During the event, professionals from each participating organisation shared real-life cyberfraud cases and practical countermeasures from their respective areas of expertise.

    Cyberfraud has been a persistent issue in Hong Kong in recent years, with victims losing increasingly large sums of money — a trend that many readers are likely already aware of. The reason this seminar was named “Let’s Secure as we Digitalise” is that frequent fraud incidents can significantly affect businesses, especially small and medium-sized enterprises (SMEs), that are planning or undergoing digital transformation. These scams may even hinder the government’s broader efforts to promote digitalisation across society.

    However, a seminar is just the beginning. For enterprises that are considering or have already embarked on the path of digital transformation, taking the first step towards cyber fraud prevention is crucial. In reality, however, many business owners struggle to identify exactly what that first step should be.

    From the perspective of CrossWise InfoTech’s founder, the answer lies in the ancient wisdom of Sun Tzu’s “Art of War“: “Know yourself and know your enemy, and you will never be in peril.“

    Since we are already aware of the various fraud tactics used online, it makes sense to evaluate whether our own defences are strong enough.

    Therefore, the right first step is to conduct a cybersecurity audit — a comprehensive assessment of an enterprise’s cyber defence capabilities based on industry standards. Through such an audit, companies can identify the risks they face in daily operations, with particular focus on areas most vulnerable to cyberfraud or other cyber threats. Based on the audit findings, and taking into account the company’s risk tolerance and financial resources, appropriate cybersecurity management measures can be designed and implemented, along with necessary technical safeguards.

    CrossWise InfoTech Limited offers customised cybersecurity audit services tailored specifically for SMEs and organisations across different sectors. Leveraging the founder’s extensive experience in enterprise IT governance and threat response, we provide actionable and effective recommendations to help clients strengthen their cyber resilience.

    At its core, cybersecurity is a battlefield of attack and defence. Thus, age-old military wisdom like that found in Sun Tzu’s “Art of War” remains highly relevant today. Both cybersecurity professionals and business leaders would benefit from reflecting deeply on these timeless strategies.

    • HKDPO: Hong Kong Digital Policy Office
    • CSTCB/HKPF: Cyber Security and Technology Crime Bureau (CSTCB) of Hong Kong Police Force (HKPF)
    • HKCERT: Hong Kong Computer Emergency Response Team Coordination Centre by HKPC
    • HKCS: Hong Kong Computer Society

    Photo of this article is shot by CrossWise InfoTech Limited.

  • CrossWise InfoTech Limited effectively uses AI technology to accomplish information system audit

    CrossWise InfoTech Limited effectively uses AI technology to accomplish information system audit

    Traditional information system audit processes require the collaboration of multiple technical personnel with different roles, and the size of the audit team can dramatically expand as the scale of the audited entity grows. This not only incurs high direct costs but also leads to hidden costs such as collaborative communication, affecting efficiency.

    CrossWise InfoTech Limited has effectively utilized AI technology in the information system audit process, particularly in the initial organization of audit working papers, logical analysis, compliance analysis, and risk point identification. These specific tasks have seen efficiency improvements by several dozen times, allowing audits to be completed in an extremely short period. In addition to essential conclusions like overall assessments, it has also effectively identified high-risk vulnerabilities in the information systems of the audited entities, which is of great significance.

    From the execution process of this audit, CrossWise InfoTech Limited believes that human application of AI technology exists at three levels: being able to operate AI to obtain results; being able to judge the correctness of the results provided by AI; and being able to control AI to achieve correct results.

    These three levels determine whether AI is a partner or a burden in human work.

    In fact, subjective initiative remains with humans.

    * Article photo is shot by CrossWise.