As the retail sector embraces agentic commerce, AI agents are increasingly facilitating tasks like browsing and purchasing on behalf of users. However, these advancements pose security challenges, as malicious actors exploit similar technologies for fraudulent activities.
This Black Friday, the primary concern is not just the proliferation of bots but discerning genuine agent-driven transactions from harmful automated interactions.
Identity verification in an automated era
Retailers need to be vigilant as paths for account creation and login are exploited by helpful AI agents
Automated interactions complicate identity verification at the account level, where the distinction between legitimate users and malicious entities is crucial. With rising automation, retailers need to be vigilant as paths for account creation and login are exploited both by helpful AI agents and harmful bots.
According to a 2025 assessment, 64% of retailers remain at risk of fake account creation, and over half face account takeover threats due to insufficient login protection.
Security holes in retail platforms
Persistent vulnerabilities have resulted in increased incidences of stolen accounts and drained gift cards, putting real shoppers at a disadvantage during peak shopping times such as the holiday season.
DataDome Advanced Threat Research tested security measures at 11 major e-commerce sites and found significant vulnerabilities to bot-driven account abuses.
Fake account creation and login protection
DataDome's findings highlighted the ease of fake account creation, with 64% of retailers at risk
DataDome's findings highlighted the ease of fake account creation, with 64% of retailers at risk. Additionally, 73% of platforms accept disposable emails, facilitating the creation of unlimited fake accounts. Only 27% effectively block bot-driven account creation, while 36% lack multi-factor authentication (MFA), leaving them vulnerable.
Concerns also extend to login protection, with 82% of retailers permitting automated login attempts without challenge and 64% lacking account lockout controls, making them susceptible to credential stuffing attacks.
Potential risks and implications
Mass fake account creation remains a major threat as the holiday shopping season approaches. Attackers utilise disposable emails and other techniques to generate numerous accounts, bypassing verification processes.
These accounts are then used to exploit purchase limits and promotions. Credential stuffing poses another significant risk, with attackers quietly testing stolen credentials at scale. As AI agents become adept at mimicking human interactions, the risk of account takeovers increases.
Gartner predicts that 90% of organisations permitting credential sharing with AI agents will experience a tripling of account takeover incidents by 2028. Retailers face the challenge of balancing user convenience with robust security measures to protect against such fraud.
Strategic recommendations
Deploying a robust bot control solution can also help identify and block refined, malicious traffic
Urgent action is needed before Black Friday. Retailers can strengthen security by blocking disposable email domains, implementing email normalisation to prevent multiple account abuses, and enforcing account lockouts after repeated failed login attempts.
Deploying a robust bot management solution can also help identify and block sophisticated, malicious traffic.
Industry-wide vulnerability
The e-commerce industry exhibits concerning levels of vulnerability to automated account abuses, with 64% of platforms not meeting basic security standards. In some cases, 18% of retailers lack even the most fundamental protections.
As Black Friday 2025 approaches, the potential for widespread fraud looms large, with risks ranging from fake accounts to large-scale account takeovers.
However, by addressing critical vulnerabilities within a short timeframe, retailers can safeguard their revenue and maintain consumer trust, staying ahead of AI-driven threats in this crucial sales period.
As agentic commerce takes hold, retailers are entering a new phase of automation, where AI agents act on behalf of real users to browse, compare, and buy. But these same capabilities can be weaponised by fraudsters.
This Black Friday, the challenge isn’t stopping bots; it’s distinguishing between legitimate agent-driven interactions and malicious automation designed to mimic them.
Identity amid rising automation
That distinction matters most at the account layer, where retailers must verify identity amid rising automation. Legitimate agents assisting users and malicious bots probing for vulnerabilities follow similar account creation and login paths.
Against this backdrop, the 2025 assessment shows that 64% of retailers remain vulnerable to fake account creation, and more than half are exposed to account takeover attacks due to weak login protections.
Stolen accounts, drained gift cards, and real shoppers
What’s more, many of the same vulnerabilities observed last year remain unaddressed.
The result? More stolen accounts, drained gift cards, and real shoppers forced to battle bots for this year’s hottest gifts — right in the middle of the holiday rush.
Key findings of DataDome Advanced Threat Research
Using open-source bot frameworks with minimal configuration, DataDome Advanced Threat Research conducted security tests across 11 major e-commerce sites to evaluate how well these platforms protect against automated account abuse. The results show widespread vulnerabilities that leave retailers exposed.
Fake account creation remains alarmingly easy
- 64% of retailers are vulnerable to mass fake account creation
- 73% accept disposable emails, allowing attackers to spin up unlimited accounts using temporary inboxes
- Only 27% of assessed retailers implement effective bot detection that successfully blocks automated account creation
- 36% of retailers have no MFA in place, leaving account creation flows dangerously open
Login protection remains weak
- 82% allow automated login attempts without challenge
- 64% have no account lockout controls, exposing them to credential stuffing attacks
These weak points provide ideal conditions for AI-driven attackers to scale their operations without being flagged; executing targeted login attempts, spinning up fake accounts, and interacting with security flows more like humans than bots.
Implications & risks
- Mass fake account creation: Fake account creation remains the most widespread and damaging threat leading up to Black Friday. Attackers use disposable email domains and simple aliasing techniques (like Gmail’s “dot” and “plus” tricks) to generate hundreds of accounts from a single inbox. Combined with automation and now AI agents that simulate real user input, these fake accounts are created at scale and often pass verification unnoticed. Once created, these accounts are used to bypass purchase limits, hoard high-demand inventory, and repeatedly redeem promotions or referral codes. The financial damage can be substantial; retailers stand to lose $50,000 – $500,000 per campaign to fraudulent promotions and resale-driven inventory grabs.
- Credential stuffing & account takeover: Credential stuffing remains a high-impact, low-visibility threat. With 55% of retailers failing to enforce account lockout or detect bot logins, attackers can quietly test stolen credentials at scale. AI agents heighten this risk, as they adapt login attempts based on platform responses, avoiding detection and increasing takeover success rates. Once inside, fraudsters exploit stored payment data, loyalty points, and user trust.
- The new risk of credential sharing in agentic commerce: According to Gartner, 90% of organisations that allow users to share credentials with AI agents will experience three times more account takeover incidents by 2028. Yet 36% of U.S. adults already say they’re interested in letting an AI agent shop or transact on their behalf. That tension—between convenience and control—will define the next wave of fraud risk. Retailers need to decide not only how to detect stolen credentials, but how to safely enable trusted agent access without opening the door to large-scale credential abuse.
- Disposable emails to bypass MFA: With 73% accepting disposable email domains, these platforms allow attackers to fully bypass MFA protections using throwaway addresses that are easy to automate and verify. The result is a false sense of security: accounts appear protected, but in practice, they’re wide open to mass fake account creation.
Recommendations: Fast fixes before Black Friday
Retailers still have time to close the most critical gaps before traffic surges. To mitigate the above risks, retailers can take steps to enhance their security posture:
- Block disposable email domains. This single change can reduce fake account creation by up to 80-90%.
- Implement email normalisation. Removing “dot” and “plus” variations from Gmail addresses can cut multi-account abuse by as much as 70%.
- Implement account lockout: After repeated failed login attempts, account lockout is essential to stop credential stuffing attacks.
- Implement the disallow directives in robots.txt and deploy a robust bot management solution to actively detect and block sophisticated, malicious traffic from AI agents.
Vulnerable to automated account abuse
The e-commerce industry shows a concerning trend: while a handful of pioneering retailers have implemented sophisticated, multi-layer defences, the majority remain vulnerable to automated account abuse, mass fake account creation, and credential stuffing attacks.
The assessment revealed that 64% of platforms fall short of baseline protections, and 18% are so exposed they lack even the most basic safeguards.
Conclusion
Black Friday 2025 carries a high risk of widespread fraud, ranging from hundreds of thousands of fake accounts to large-scale account takeovers.
The good news is that most critical vulnerabilities can be resolved within 24 to 48 hours; retailers who act now will be in a strong position to protect revenue, preserve trust, and stay one step ahead of AI-driven threats during the year’s most important sales window.
