AI-written edition with a fictional anchor
Here is today's evening report. Fri, Oct 23, 18:14 HKT
13:08 HKT ConnectWise compromise leaves access open where supplier credentials still work Kai Tanner explains why the ConnectWise compromise reaches customers wherever their systems still accept the supplier's credentials. read more
Kai Tanner, CYBER INTEL
AI-authored analysis · Fictional correspondent, real sourcesAn agent can accelerate the work; the permissions attached to an accepted credential determine what that work can reach. For an administrator reading about the ConnectWise compromise, that puts a concrete check ahead of the debate about automation: which accounts used by the supplier can still sign in, and what can they do?
The Record reports that a Chinese government-aligned group used automated tools in the attack on ConnectWise, an IT management software company. Microsoft’s account describes agents scaling supply chain compromise and credential abuse across enterprise environments. Neither description establishes which downstream accounts remain usable. That requires evidence from the receiving system.
The supplier has an incident to investigate. The customer has an access decision.
A successful sign-in records that the receiving system accepted the credential. Establishing who used it requires more evidence. An account can retain its familiar name and permissions while the circumstances in which it is being used have changed.
That leaves the administrator with two records to reconcile: the supplier’s account of the compromise and the customer’s own record of accepted access. The relevant evidence is the account, its permissions and its activity. The presence of an AI agent does not answer any of those questions.
Disabling a compromised supplier account or revoking its credentials would close that route. Where the downstream system continues to accept them, whoever holds those credentials retains the access that system grants.
Sources13:05 HKT LiteLLM breach shows model tests offer no assurance over credential safety Rachel Lam examines what the LiteLLM breach means for Hong Kong financial firms assessing the software connecting them to AI. read more
Rachel Lam, FINANCE & RISK
AI-authored analysis · Fictional correspondent, real sourcesA firm can evaluate a model’s answers while the software connecting to it steals the credentials. Attackers compromised LiteLLM’s PyPI package and published malicious Docker images carrying a payload that steals credentials, secrets and environment variables (the model can pass its tests while the connection fails its own).
Kai Tanner’s filing today identifies both delivery routes. For the people approving an AI supplier and its connection inside a Hong Kong licensed firm, that puts a compromised software component directly into the approval decision. A satisfactory answer from the model establishes nothing about whether the application has kept its credentials.
So the SFC’s Intermediaries Division has an existing instrument that reaches this decision. Its November 12, 2024 circular on generative AI language models covers senior management responsibility, model risk, cybersecurity and data risk, and third-party provider risk. The supplier assessment reaches the software through which the firm uses the model.
Aya Nakamura’s filing today reports Uber’s rollout of ChatGPT Work to 30,000 employees. That supplies the scale of an enterprise adoption decision; the LiteLLM incident supplies a concrete failure in connection software. Neither report establishes that Uber used the compromised software.
For a licensed firm putting client-facing investment research into service, the regulatory consequence is specific. The SFC circular identifies that use as high risk and notifiable under the Securities and Futures (Licensing and Registration) (Information) Rules.
Sources
Continue reading
State Investor Buying Helps Chinese Stocks Reverse Losses
China's market recovery shows the influence of state investors, even as a wider Asian technology selloff exposes the limits of that support.
Chinese stocks reversed losses as money flowed faster into exchange-traded funds favoured by state investors, Bloomberg reports. The recovery followed Thursday's stock market selloff and came as Asian markets faced further selling in AI shares. Buyers associated with China's 'national team' helped steady the domestic market. That changed the day's direction for Chinese stocks without establishing that investors across the region had recovered their appetite for technology shares. State support can produce a recovery in prices before there is a recovery in confidence.
The sequence matters. Thursday's selloff had already inflicted one of hedge funds' broadest daily declines since Covid, according to Goldman Sachs's prime brokerage desk, as reported by Bloomberg. By Friday, investors in Asia were seeking safety as AI stocks fell. Then accelerating inflows into the funds favoured by Chinese state investors helped Chinese shares reverse their losses. These reports describe different parts of the market, rather than a single measure of its health. A rebound in Chinese stocks can therefore coexist with substantial losses elsewhere. The national team can buy shares; it cannot purchase a regional mood.
Our correspondents have separate reporting on the technology businesses behind the investment debate. Kai Tanner's Cyber Intel desk examines the reported ConnectWise compromise and establishes why customer systems' acceptance of supplier credentials matters after an attack. Rachel Lam's Finance & Risk desk takes the LiteLLM breach into Hong Kong's regulated financial sector, explaining why testing a model's answers provides no assurance that its connection software protects credentials. Neither filing explains today's share prices. Both supply detail that a market valuation cannot provide about how the technology actually operates.
The market's conclusion is narrower than the rebound might suggest. Chinese shares found support from buying associated with state investors during a wider retreat from risk. That support mattered to prices. It does not, by itself, establish renewed private demand or restore the expectations that made technology shares attractive before the selling began.
All episodes and how to subscribe →