AI Agents Lack Legal Identity as Industry Races to Build Credential Standards

·4 views

AI agents that execute financial and other high-stakes tasks currently operate without verifiable legal identities, relying on shared logins and overly permissive API keys. Developers working on the problem have built a passport-style credentialing system running on public blockchain infrastructure and open credential standards, with a draft specification submitted to the IETF. The system defines an agent's origin, operator liability, permitted actions, and validity period, with sub-agents inheriting only narrower authority than the parent. Despite functional infrastructure existing by mid-2026, adoption remains negligible because most platforms do not yet require agents to present credentials before transacting. Regulatory deadlines, particularly from the EU's forthcoming rules on high-risk AI, are expected to force the issue and finally close the gap between available standards and real-world enforcement.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Tailscale SSH Flaw Allowed Unauthorized Root Access via Argument Handling Bug

Tailscale has disclosed a security vulnerability tracked as TS-2026-009 affecting its SSH feature. The flaw stemmed from insecure argument handling, which could allow an attacker to gain root-level access on affected systems. The bulletin was published on Tailscale's official security advisories page. Users relying on Tailscale SSH are advised to review the advisory and apply any recommended mitigations or updates promptly.

0 comments Read more at Hacker News

ProgrammingDEV Community ·

82% of Ad Campaign MCP Servers Cover Only One Platform, Survey Finds

A survey of ad-campaign MCP servers, drawing from the official MCP registry and the community awesome-mcp-servers directory, identified 17 distinct tools available as of July 2026. The analysis found that 14 of those 17 servers — roughly 82% — support only a single advertising platform such as Meta, Google, or TikTok. Just three servers cover two or more platforms, and only one spans six platforms. The fragmented landscape means advertisers relying on AI recommendations often end up managing multiple separate installs, credential stores, and maintenance surfaces. The survey was conducted and published by Adspirer, which itself operates a multi-platform MCP server and acknowledges a conflict of interest in the findings.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

AI Eliminates 90,000 US Jobs While Blocking Entry-Level Candidates From New Ones

A June 2025 Nikkei Asia report found that top US university tech graduates collectively sent out 8,000 resumes with almost no responses, while TechCrunch data showed over 90,000 US positions had been explicitly eliminated due to AI before May 2026. An MIT NANDA report found that 95% of enterprise AI deployments show no profit impact, but the 5% that do generate returns almost exclusively use AI to replace repetitive entry-level tasks such as data processing, code review, and content editing, with savings reflected directly in headcount cuts. The roles most affected include junior technical documentation, customer support, content editing, and entry-level analyst positions in finance and market research. A particularly stark dynamic has emerged where AI tools are being used to screen resumes and conduct first-round interviews, effectively rejecting candidates for roles that AI itself has already made redundant. However, demand remains strong for workers who can design AI evaluation frameworks, manage multi-agent systems, and perform tasks requiring complex human judgment that current AI agents cannot reliably handle alone.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

CSPM, CWPP and CNAPP: How Three Cloud Security Tools Differ and Work Together

Cloud environments on AWS, Azure, and GCP commonly face three distinct security challenges simultaneously: misconfigured infrastructure, vulnerable workloads, and lack of visibility into how these risks connect. CSPM (Cloud Security Posture Management) continuously scans cloud configurations against benchmarks like CIS and PCI-DSS to detect errors such as public storage buckets or over-privileged IAM roles. CWPP (Cloud Workload Protection Platform) focuses on protecting what runs inside the cloud — virtual machines, containers, and serverless functions — checking for vulnerabilities within active workloads. CNAPP (Cloud-Native Application Protection Platform), a term coined by Gartner around 2019, was designed to unify CSPM, CWPP, and additional security layers into a single platform offering end-to-end risk visibility from code to runtime. Industry research consistently identifies misconfiguration, rather than software vulnerability exploitation, as the leading cause of cloud security incidents.

0 comments Read more at DEV Community