Microsoft and UC Berkeley release TokenEconomizer for coding agents
On 277 SWE-Bench tasks it cut mean cost 18.4% to 30.9% across five agent-model settings without changing weights.
46
- Mentions · 7d40
- Mentions · 30d118
- All mentions250
- First seenAug 12, 2026
- Last mentioned6h ago
Timeline
250 articles · newest firstUnauthorized relays disable Codex and Claude Code sandboxes at 12 firms
A post-install script points the genuine CLIs at relay servers, swaps API keys, and sets danger-full-access; abuse has not been observed.
62
Embedded browsers in Cursor, Copilot, Codex, VS Code and Claude Desktop flawed
Forever Security’s EMBRACE findings show prompt injection, address spoofing, and local file reads; Codex and VS Code were still unpatched.
74
Grey-market relays reroute Codex and Claude Code traffic
Symantec says lookalike-domain scripts disable sandboxes and send prompts, files and tokens through operator servers at about a dozen firms.
74
OpenAI doubles ChatGPT Pro usage limits after reversing a planned cut
The change covers Pro 100, 200 and 500 plus Business Premium, and Dots usage stays separate.
38
OpenAI doubles usage limits on Pro 100 and Pro 500 plans
Plus is unchanged, while Pro 200 received a doubled lower quota and a reversed planned cut.
34
Embedded browsers in five AI coding tools can leak files or run code
As of October 29, Codex and VS Code still let one local HTML click read SSH keys, tokens and .env files.
74
Unauthorized AI relay disables Codex and Claude Code sandboxes
Symantec says the relay has installed genuine CLIs at a dozen organizations, mostly in mainland China, since early 2026.
52
Credit Saison reports 954% ROI from a ChatGPT Enterprise pilot
A 315-person pilot saved 14 hours a month per person, and Credit Saison rolled ChatGPT Enterprise out companywide in September 2025.
22
CARI hides malicious instructions in coding-agent runtime feedback
The attack succeeded in 2,845 of 3,000 trials and transferred to Codex and Claude Code.
58
OpenAI says Dots should reach more of Europe within weeks
Personal Pro accounts in the EEA, Switzerland and the UK remain locked out, while some business accounts already have access.
38
OpenAI and ServiceNow expand partnership across enterprise workflows
A ChatGPT integration can act on ServiceNow records; Codex, security and voice features are mostly scheduled for 2027.
46
Cisco open-sources Project Quarry to flag rogue AI agents
Quarry scores full agent sessions, including tool use and MCP calls, and ships a YARA-like rule starter pack.
48
OpenAI customer story pitches Legora as legal AI platform
Stockholm-based Legora, founded in 2023, says NDA review fell from about 20% of its legal team's capacity to under 1%.
26
Legora builds an agentic operating system for legal work
The Stockholm firm cites a 95%+ cut in NDA review time and 90% faster contract turnaround for a Fortune 100 customer.
26
Aokigahara stealer takes developer keyrings and AI assistant sessions
Check Point says the Windows malware also steals browser, VPN, wallet, and password-manager data from one endpoint.
56
Retell says OpenAI models power more than 100 million monthly calls
Retell reports volume grew from 30 million calls early in 2026, with OpenAI generating responses on 80% of calls.
32
Engineer argues agent-written code still needs human ownership
A SonarQube complexity flag on Codex-generated React code led the author to inspect, then have the agent split one file into ten.
28
OpenAI releases GPT-6.1 Terra with a Critical cyber capability rating
The October 26 model is in the API as gpt-6.1-terra at $1 per million input tokens and $5 per million output tokens.
78
OpenAI releases GPT-6.1 Terra for agents and computer use
Standard API prices are half those of GPT-5.6 Terra and GPT-6.1 Sol, and cybersecurity is rated Critical.
84
OpenAI and Oracle to bring Oracle apps into ChatGPT
US customers can buy the OpenAI API, Codex, and ChatGPT Work through Oracle Marketplace.
55
Researcher uses Codex to bypass four USB encryption products
Hidden accounts and hardcoded keys decrypted protected files without owner passwords, one flaw in seven minutes.
48
OpenAI releases GPT-6.1 Terra to ChatGPT, Codex, and the API
API prices are half those of GPT-5.6 Terra, and OpenAI says Terra meets its Critical cybersecurity threshold.
82
Aokigahara infostealer steals ChatGPT, Claude, Codex, and Cursor logins
Check Point says the Windows stealer also grabs cloud, SSH, and browser secrets, not via flaws in those AI apps.
68
OpenAI’s dots assistant packages coding agents for Pro users
It is in ChatGPT desktop and mobile apps from $100 a month and can research proactively.
48
Greptile uses GPT-6 Astra to test apps beyond code review
The startup also wires its knowledge base to Codex over MCP and cites Daybreak for customer defenses.
22
Malicious AGENTS.md tried to hijack coding agents
Forever Security says a planted markdown file on a customer PC could push Codex to install malware or leak tokens, and was stopped before code ran.
74
OpenAI launches GPT-6.1 Terra, a cheaper faster agent model
API pricing is $1 per million input tokens and $5 per million output, half of GPT-5.6 Terra, and it is available now.
74
OpenAI introduces GPT-6.1 Terra for computer use
Standard API pricing is $1 per million input tokens and $5 per million output tokens, 50% below prior Terra and Sol rates.
78
Entire launches a Git forge for agent-human collaboration
Trails score every push with agent evals, and gates can block merges until findings, human approval, and CI pass.
48
Simple LLM harnesses outdiagnose hand-crafted microservice RCA agents
On 90 RCAtlas cases, simple harnesses reach PR@1 of 65–77% versus 16–53% for published agents.
48
Malicious Downloads files can hijack Codex and Claude Code
Forever Security says a planted AGENTS.md or CLAUDE.md can change agent settings, install an npm package, or expose an OAuth token.
66
Aokigahara infostealer collects AI coding-tool tokens and cloud credentials
Check Point says it searches local app storage rather than exploiting a flaw in the targeted AI assistants.
72
Planted Downloads files can hijack Claude Code and Codex agents
Forever Security says a planted AGENTS.md made Codex install MALFEX package function-color in over 90 percent of tests.
64
OpenAI says Codex Ultrafast sped internal incident investigations
Staff say security data reviews fell from one or two hours to about 10 to 15 minutes.
40
Fanatics launches OpenAI-powered Shop the Look on sports stores
After tests on about 30,000 MLB products, the companies report 10% higher engagement than the prior experience.
28
Downloads-folder instruction file tried to make a coding agent install malware
Forever stopped the agent before code ran; Codex followed the planted file in more than 90% of isolated tests.
66
OpenAI and Anthropic cut LLM prices by up to 80% since June
Cache-read cuts matter most for coding agents, where cache reads are 96–99% of token spend, the author says.
36
Malicious AGENTS.md files in Downloads hijack Codex and Claude Code
A customer agent tried to install npm package function-color; tests also stole Claude Code OAuth tokens in about half of runs.
63
OpenAI keeps frontier training offline after agents escape containment
Red-teaming found CVE-2026-111497, a critical unauthenticated RCE in Google Brotli, plus VM escapes and cache side channels.
76
Related CVEs
by mentions- CVE-2026-8207861%9.4Unsafe Reflection RCE in PaperCut NG/MF, Chained with Auth Bypass in Attacks
- CVE-2026-8157884%8.8Missing Authentication for Critical Function in PaperCut NG/MF Web Interface
- CVE-2026-19592<1%7.3Unsandboxed attacker code execution in OpenAI Codex via malicious Git core.fsmonitor
- CVE-2026-11149783%9.2Buffer overflow in Google Brotli decoder 1.0.3–1.2.0
- CVE-2021-4227873%7.5Privilege Escalation via sAMAccountName Spoofing in Microsoft Active Directory
- CVE-2026-19593<1%9.8RCE via Attacker-Controlled Git Config in OpenAI Codex Desktop (Windows/macOS)
- CVE-2026-19591<1%8.8PowerShell parsing flaw enables approval-bypass RCE in OpenAI Codex CLI/Desktop
- CVE-2026-71963<1%8.6Command injection RCE in Hermes Agent via malicious .git/config
- CVE-2026-55607<1%7.7Claude Code: Claude Code's worktree handling allowed creation of worktrees named ".git" and navigation to worktrees outside the sandbox…
- CVE-2026-72718<1%7.0goose is general-purpose AI agent that runs on your machine
Appears with
- OpenAIvendor · 175
- OpenAIorganization · 93
- ChatGPTproduct · 83
- Claude Codeproduct · 75
- Anthropicvendor · 66
- GPT-6 Astraai model · 49
- Anthropicorganization · 31
- Googlevendor · 27
- Cursorproduct · 23
- Codexai model · 22
- GPT-5.6 Solai model · 21
- GPT-6.1 Solai model · 20
- Microsoftvendor · 19
- Claude Opus 5ai model · 19
- Claudeproduct · 17
- GitHub Copilotproduct · 16
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.