Internet Engineering Task Force
PulseAugur coverage of Internet Engineering Task Force — every cluster mentioning Internet Engineering Task Force across labs, papers, and developer communities, ranked by signal.
8 day(s) with sentiment data
-
Cloudflare flags AI agent identity challenges for websites · 4 sources tracked
Cloudflare has highlighted significant challenges posed by AI agents to website operators, particularly concerning identity and accountability. The company argues that websites bear the cost for nearly all issues arisin…
-
Google, OpenAI, Microsoft, Anthropic to honor shared AI training opt-out · 1 source tracked
Major AI companies including Google, OpenAI, Microsoft, and Anthropic have announced plans to honor a shared robots.txt signal for AI training opt-outs. This initiative, discussed at the Internet Engineering Task Force,…
-
Anthropic technical lead proposes AI policy token at IETF
Kevin Kelley, a technical lead at Anthropic, has proposed a "policy token" signal within the Internet Engineering Task Force. This proposal aims to standardize how AI systems communicate their policy compliance.
-
AI giants agree on new standard to control web scraping for training
Major AI companies including OpenAI, Google, Microsoft, and Anthropic have agreed to adopt a new standard for website scraping. This initiative, proposed by Anthropic's Kevin Kelley and to be integrated into the robots.…
-
New llms.txt format aims to guide AI crawlers to key website pages
The llms.txt format is a community-driven proposal for websites to provide AI tools with a curated list of important pages. This Markdown file, typically located at the root of a website, includes a site title, a summar…
-
MCP protocol detailed for Python security testing and Java server development
This article explores testing the security boundaries of an MCP proxy, focusing on how it handles OAuth tokens. It proposes a Python-based policy fixture to make these trust boundaries testable, emphasizing the separati…
-
Typo-squatting attacks exploit Unicode complexities, evading browser defenses
Typo-squatting attacks remain a significant threat, particularly due to the complexities of Unicode characters in domain names. While browsers like Chromium have advanced defenses, specific Cyrillic characters can still…
-
AI agent evidence draft proposes minimum standards for verifiable actions
A working draft called "Proof of Control" is proposing a minimum standard for AI agent evidence, aiming to ensure that systems provide verifiable records of agent actions. The draft, open for public comment until Octobe…
-
AI agents to log skipped tasks with new execution receipt spec
A new specification for AI agent execution receipts, AER-1, has been proposed to address the critical issue of "omission failures" where agents silently skip tasks or verifications. Unlike traditional logs that only rec…
-
DNS faces obsolescence with new security protocols and industry backing
The Domain Name System (DNS) is facing obsolescence due to emerging technologies that offer enhanced security and efficiency. Protocols like DNS over HTTPS (DoH) and DNS over TLS (DoT) are gaining traction, providing en…
-
MCP Cross App Access Explained: Enterprise IdP Controls Server Access
The MCP Cross App Access (XAA) extension enables enterprise identity providers to manage user access to MCP servers. This process involves a three-step HTTP exchange where a client obtains an ID token from the IdP, exch…
-
Zamboanga introduces verifiable AI receipts to prove job completion
Zamboanga has introduced a verifiable execution receipt system, dubbed "Zambo receipts," designed to provide irrefutable proof of AI job completion. These receipts are generated by the execution layer itself at the mome…
-
AI agent receipts: Integrity, trustlessness, and veracity claims analyzed
A new approach to auditing AI agents involves signing their actions, creating a verifiable receipt. This method, supported by open protocols and tools like Microsoft's Agent Governance Toolkit, offers three distinct gua…
-
Idempotency test ensures AI agents don't create duplicate orders
This article details the third leg of a parity test for API protocols, focusing on idempotency, which is crucial for AI agents that automatically retry operations. The author demonstrates how a lack of idempotency can l…
-
Agent economy grapples with settlement gap despite routing and payment advances
The agent economy faces a critical challenge in its settlement layer, despite advancements in routing and payment mechanisms. While protocols like AGTP-COMMERCE handle agent discovery and verification, and payment stand…
-
Cloudflare enables opt-out of AI training while preserving search indexing
Cloudflare is introducing a new feature that allows website owners to differentiate between content indexing for search engines and content usage for AI training. Starting September 15, 2026, websites can opt out of AI …
-
Developer adds cryptographic receipts to MCP tool calls for enhanced security
A developer has implemented cryptographic receipts for MCP (Model Context Protocol) tool calls to enhance security and verifiability. This technique, integrated into the ccs-mcp-server library, adds approximately 20 lin…
-
New tool secures MCP tool calls with 3-line CI integration
A developer has released a new tool, `ccs-lint-action`, designed to secure Model Context Protocol (MCP) tool calls by detecting and flagging unsigned or improperly nested integrity fields within JSON responses. This act…
-
Tencent study finds DeepSeek Harness vulnerable to prompt injection; Correctover offers solution
A security assessment by Tencent's AI-Infra-Guard team revealed that DeepSeek Harness is vulnerable to indirect prompt injection attacks, with success rates ranging from 17% to 25.5% across various channels. The study c…
-
IETF Discusses AI's Impact on Internet Standards at IETF126
The Internet Engineering Task Force (IETF) is considering the implications of artificial intelligence (AI) for standardization processes. Discussions at IETF126 focused on how AI technologies might influence or require …