Loading story…
Multiple outlets reported growing adoption of agentic AI—systems that can autonomously execute multi-step workflows—across commerce and the public sector, raising new security, fraud, and accountability concerns. In e-commerce, AI “shopping” agents are increasingly able to initiate purchases and interact with merchant sites; reporting highlighted merchant complaints that automated agents can scrape product data without consent, generate inaccurate listings or images, and create fulfillment/pricing errors that damage customer trust while leaving merchants to absorb the operational and reputational impact.
In financial services, commentary on stablecoin payment fraud argued that because many crypto transactions are effectively irreversible, fraud controls must shift “left” to pre-authorization blocking using real-time AI models tuned for high recall, supported by layered defenses and use of open blockchain telemetry. Separately, state and local government IT leaders are exploring agentic AI to improve productivity and citizen services, implying a need for stronger governance over autonomous actions, auditability, and risk controls as these systems move from pilots into operational workflows. One reference was a vendor/platform announcement and another was a broad societal-risk article on AI (deepfakes, opacity, labor disruption) that did not materially add to the specific agentic-payments/fraud/government adoption thread.
Track how attackers are adapting to this technology.
5 events from the most recent confirmed update back to the earliest known activity.
In a March 5 interview, Donald Kossmann said enterprises should not assume an AI agent's authorized transaction reflects true user intent and warned of risks such as intent drift, decision-integrity failures, and agent-to-agent marketplace abuse. He recommended granular and revocable permissions, detailed audit logs, human override, and post-transaction evidence capture before connecting agents to procurement systems or payment methods.
In a March 4 interview, Sean Ren argued that stablecoin fraud prevention must rely on AI operating in milliseconds before authorization because transactions are effectively irreversible. He recommended layered defenses optimized for high recall and said open blockchain transaction data can strengthen fraud-detection models.
Reporting on the rise of AI-driven shopping and payments said agentic commerce is introducing security, fraud, and liability problems for merchants and platforms. The coverage cited concerns over unauthorized scraping, incorrect orders, customer-trust damage, and unclear responsibility when AI-initiated transactions go wrong.
A National Association of State CIOs report described agentic AI as an emerging frontier for state IT and warned that agents accessing multiple systems and databases expand the security and governance risk surface. The report recommended using frameworks such as NIST's AI guidance as a starting point for oversight.
U.S. state governments started cautiously exploring agentic AI beyond generative AI, with about eight states reporting initial use. Examples include Alaska evaluating agentic AI in the myAlaska portal, Tennessee considering agentic features in next-generation ERP, and Virginia piloting a tool to speed regulation review.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
Follow how adversaries are adapting to this technology, and where it touches your stack today.
6 references tracked. Mallory keeps watching after this page renders.
datainnovation.org
Open sourcehelpnetsecurity.com
Open sourcecio.com
Open sourcebankinfosecurity.com
Open sourcebankinfosecurity.com
Open sourcegovinfosecurity.com
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Banks and payment providers are warning that **deepfakes** and agentic AI are increasing fraud pressure on digital payments, particularly where identity verification, authorization, and anti-money-laundering controls depend on signals that can be spoofed. Deloitte highlighted rising concern that AI-generated voice and video can be used to impersonate customers and defeat banking verification workflows, while other industry reporting noted warnings that deepfakes could bypass AML and identity checks and expose institutions to higher account takeover and payment fraud risk. At the same time, firms are adapting payment infrastructure for AI-driven commerce rather than allowing autonomous models to move money without guardrails. Industry analysis said payments remain difficult for AI agents because settlement requires licensed intermediaries, deterministic execution, regulatory compliance, and clear liability for fraud and errors. Providers are responding with agent-focused controls such as wallet-based authorization, token protections, and abuse detection, reflecting a broader shift toward **constrained, supervised AI payments** instead of fully autonomous transaction execution.
Aug 4, 2026
Cybersecurity and fraud-risk leaders are urging stronger, enforceable security controls as the U.S. **Department of the Treasury** develops AI cybersecurity and risk-management guidance for the financial sector. Reported expectations for the forthcoming materials include coverage of **governance**, **data practices**, **fraud**, **transparency**, and **digital identity**, but analysts warned that non-mandatory “principles” risk being ineffective as **AI-enabled attacks** increase. Recommended guardrails include mandatory **adversarial testing**, maintaining an **AI inventory**, **model monitoring**, and **real-time identity validation**, with specific concern for threats such as **data poisoning** of training pipelines. In parallel, financial institutions deploying **agentic AI** (systems that can initiate payments, approve transactions, or freeze accounts) are being warned that traditional authentication models assume a human user and do not adequately address autonomous or semi-autonomous agents. Fraud and AML specialists are calling for **managed, revocable digital identities** for AI agents—ideally with **cryptographic proof**—and **continuous trust** approaches that validate not just identity but **delegated authority**, detect “scope creep,” and preserve traceability to both the authorizing human and the executing agent.
Jun 29, 2026
Reporting on a new MIT-led survey of 30 widely used **agentic AI** systems describes a security posture marked by **limited risk disclosure**, weak transparency, and inconsistent safety protocols, with researchers warning it is difficult to enumerate failure modes when developers do not document capabilities and controls. The coverage also points to recent attention around the open-source agent framework *OpenClaw*, citing reported security flaws that could enable **PC hijacking** when agents are granted broad permissions (e.g., to operate email and other user workflows), and includes vendor responses from Perplexity, OpenAI, and IBM. Separate industry analysis highlights how increasingly autonomous agents—especially those able to **initiate transactions**—compress detection windows for abuse and complicate attribution and liability, particularly in crypto and cross-chain contexts where funds can move in seconds. A vendor blog argues that accountability still ultimately rests with the humans who design, deploy, authorize, or benefit from these systems, and that governance/monitoring architecture may become central evidence in enforcement actions; it also claims 2025 illicit crypto volume reached **$158B** and that **AI-enabled scams** rose sharply year over year. Broader software-engineering commentary reinforces the trend toward AI-native development and widespread use of AI coding tools, but is largely directional and does not add specific incident or vulnerability detail beyond the general risk discussion.
Jun 30, 2026