会计师事务所会话令牌失窃风险与分层防护建议
Beyond the login: Session token theft and the accounting firm
会计师事务所的会话令牌可被攻击者直接盗取,使其无需破解密码或绕过第二因素便能继承已验证会话。SpyCloud 报告称仅在 2025 年就追回了 8.6 billion stolen cookies,相关风险涉及事务所掌握的社会安全号、银行资料、工资记录和报税表。

beebright - stock.adobe.com
For years, accounting firms have been told that strong passwords and multifactor authentication are the foundation of cybersecurity. That advice still holds, but cybercriminals have moved on. Today's attackers increasingly skip the login screen altogether and go after the session token, the digital credential a system hands out once a user has successfully signed in.
Processing Content
A session token is what keeps a staff accountant logged in on the tax platform, the general ledger, or the client portal without retyping credentials every few minutes. To the application, it is proof that the user has already been verified. An attacker who copies that token inherits the verified session itself, with no need to crack a password or defeat a second factor.
The volume is staggering. SpyCloud reported recovering 8.6 billion stolen cookies in 2025 alone. For firms entrusted with Social Security numbers, bank details, payroll records, and tax returns, that trend sits squarely within existing obligations under the FTC Safeguards Rule, IRS data security guidance, and professional confidentiality standards.
MFA guards the door, not the room
Multifactor authentication answers a single question: is this person the individual they claim to be at the moment of login? It says nothing about who is using the session an hour later. Confirming identity and protecting an active session are separate controls. Firms that rolled out MFA to meet insurer or regulatory requirements can easily assume they have addressed both.
Passkeys improve on passwords at the login step but share the same limitation once the session is open. They also bring practical headaches. Passkeys created in Apple, Google, and Microsoft ecosystems do not move freely between them, and a lost phone can lock a user out entirely—a real disruption in the middle of busy season.
Where stolen tokens come from
Most token theft traces back to a handful of familiar sources:
- Infostealer malware. Once on a laptop, often through a phishing attachment disguised as a client document, it silently harvests browser cookies and saved logins.
- Rogue browser extensions. Add-ons that promise faster PDF conversion or spreadsheet shortcuts can quietly send session data to an attacker.
- Flaws in web applications. Weaknesses such as cross-site scripting were once hard to find. AI-assisted tools now surface them in seconds, and a single flaw in a client upload portal can leak the sessions of every user connected to it.
Single sign-on amplifies the damage. One token from a central identity provider may open the tax software, the practice management system, cloud accounting platforms, and email at once, and long session lifetimes let an intruder linger. Staff who reuse firm logins for personal or outside services add another path: a breach at that third party can expose the firm session too.
Why no alarm sounds
A guessed or phished password tends to leave a trail: repeated failed logins, locked accounts, blocked connection attempts. A replayed token leaves almost nothing, because the system treats it as a legitimate, already-authenticated session. A partner reviewing returns from the office and an attacker using that partner's stolen cookie from another country can look the same in the logs.
This gap is built into how most monitoring works. Traditional SIEM tools flag deviations from known rules, and a valid session from an unfamiliar location violates none of them. Behavioral analytics help by learning what normal activity looks like, but telling a hijacked session from a real one in real time still means stitching together device logs, location, and timing, with frequent false alarms and missed events.
The first warning is usually indirect, such as an impossible-travel alert or a sudden burst of downloads from the document portal. Often, confirmation arrives only when the firm's data turns up for sale on criminal marketplaces. That is why dark web monitoring belongs in every firm's security program, and why delayed discovery can complicate incident response and notification duties.
Building layered defenses
Shortening token lifespans helps, but no single fix closes this gap. The answer is ongoing verification throughout the session, not just a strong check at the start.
To achieve that, there are several practical steps you can take today, starting with managing the browser. Firm-controlled browsers can limit how long cookies persist and block risky extensions, treating the browser as the high-value workspace it has become.
It's also important to reinforce daily discipline. Staff should never use firm credentials for personal sites, and firm email belongs only on firm-managed, monitored devices. These habits are worth writing into the firm's documented information security plan, with an experienced security partner helping to enforce them.
Another measure to consider is to adopt zero trust in identity. Shorter sessions with regular reauthentication, tokens bound to the device that received them, continuous session checks, and least-privilege access by role or engagement all shrink what a stolen token can reach.
Finally, set traps. Honeytokens, decoy credentials with no legitimate purpose, raise an immediate, high-confidence alert the moment anyone tries to use them. Mature security programs rely on this proven technique.
Structure matters more than spending
Firms that catch token theft early are distinguished less by budget than by how their security is organized.
A less mature firm typically depends on one identity provider and handles departures with a short checklist: disable the account; reclaim the license. With seasonal preparers and frequent staff turnover, that leaves delegated mailboxes, shared client folders, and lingering permissions open with no one watching them.
A mature firm connects multiple identity and cloud security tools, monitors its key SaaS applications, and offboards users across every system they touched, instead of leaving it all to a lone IT generalist.
Passwords were only ever the easiest way into an account. Attackers have found an easier one. Firms that still defend only the login are protecting the wrong asset, and their clients' financial data is what stands to be lost.

Chief technology officer
Vikram Chabra is the chief technology officer at NopalCyber, where he leads the technology strategy and solutions the company builds to protect... Read full bio

Cybersecurity analyst
Antonio Chelala is a cybersecurity analyst at NopalCyber, where he supports client security hardening, detection, and policy work. His focus includes... Read full bio
来源:Accounting Today(美国会计行业媒体) · accountingtoday.com