Symantec says new Aoki infostealer versions steal API keys and tokens for AI coding assistants, plus AWS credentials, from infected machines. The firm says threat actors now value such credentials like payment card data.
Reality
- Evidence50
- Adoption
- Not scored
- Hype gap+15
- Incentives40
- Confidence55
buildConfirmed2 publishers GPT-6-Astra, running in OpenAI's Codex, solved 49 of 50 unpublished theoretical-physics problems on the TheoPhysEval benchmark. The problems are routine research calculations, and the authors say AI still lacks the "taste" for breakthroughs.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+25
- Incentives
- Not scored
- Confidence60
AgentMail's agent tests trace most failures to tools that leave the agent guessing
AgentMail says 80% of runs finished in its tests of 25 agents, and most failures came where the tool left the agent guessing. The company also built the API and wrote the tasks, so the findings work best as a checklist for your own tools.
Publishers:agentmail.to
Reality
- Evidence50
- Adoption
- Not scored
- Hype gap+15
- Incentives60
- Confidence55
buildOne report1 publisher JetBrains says its IDE engine, benchmarked by Netflix in Claude Code and Codex, makes coding agents typically 20 to 30% more efficient
Netflix and JetBrains report that a headless IntelliJ code engine made Claude Code and Codex typically 20 to 30% more efficient, up to about 65%. Netflix wrote the daemon that keeps the index warm because JetBrains had not shipped one.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+20
- Incentives70
- Confidence50
Revenue from coding agents has grown faster than AI data center spending, a Noahpinion post says
Revenue from coding agents grew faster than hyperscaler data center spending, according to the Noahpinion blog, even as the companies began to borrow. The post says the new debt raises the stakes if that growth slows.
Reality
- Evidence35
- Adoption
- Not scored
- Hype gap
- Not scored
- Incentives
- Not scored
- Confidence40
Groundcover routes production debugging from Claude and Codex to its own agent
Groundcover opened its observability data to outside coding agents such as Claude and Codex through MCP tools and asynchronous Agent Tasks. A coding agent that cannot see past deployment can now pass a production question to a specialist agent.
Reality
- Evidence25
- Adoption
- Not scored
- Hype gap+20
- Incentives70
- Confidence55
buildOne report1 publisher Ruflo's 100-agent swarm for Claude Code starts in read-only mode with approval switched on
Cognitum One released a new Ruflo on October 13 that turns one coding assistant into a swarm of 100+ agents. The parts worth copying are the controls around the model: approval modes, cost visibility and data gating.
Reality
- Evidence25
- Adoption55
- Hype gap+30
- Incentives70
- Confidence55
buildOne report1 publisher Auditors reproduced 45-plus billing bugs across 110 AI usage tools, most often retries counted as new work
An audit of 110 AI usage tools reproduced more than 45 metering bugs, most often retries counted as new operations. The authors did not publish an overbilling rate, so teams paying per usage should check their own logs before trusting an invoice.
Reality
- Evidence45
- Adoption10
- Hype gap+20
- Incentives55
- Confidence50
buildOne report1 publisher Novamira's docs say AI abilities for WordPress should be 'never on production'
Novamira's free WordPress plugin lets Claude Code, Cursor or Codex run PHP, WP-CLI, database queries and file edits on a site. According to a dev.to review, Novamira itself says only the operator's attention stands between the agent and production.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap−10
- Incentives65
- Confidence50
REA, a toolkit that points AI coding agents at Ghidra, passed 100,000 GitHub stars in six days
REA, an open-source toolkit linking AI coding agents to Ghidra, went from under 10,000 GitHub stars on Oct. 6 to over 100,000 by Oct. 12. Stars are bookmarks. The evidence for cheaper cloning in Decrypt's account is anecdotal.
Reality
- Evidence40
- Adoption25
- Hype gap+50
- Incentives
- Not scored
- Confidence45
buildConfirmed2 publishers Maurice Heumann's Claude and Codex agents rebuilt 99% of Modern Warfare 2's functions in three months, 83% matching the original byte for byte. Heumann says the hard part was verifying the output and keeping the agents from gaming the check.
Reality
- Evidence40
- Adoption
- Not scored
- Hype gap+15
- Incentives
- Not scored
- Confidence55
Aptos rolls out AI tooling that lets coding assistants learn Move from its documentation, with an MCP server installed via one npx command
Aptos shipped an MCP server, LLMs.txt feeds and agent skills so AI coding assistants can read Move documentation directly, cryptobriefing.com reported. The only quality control it describes is a set of skills for audits and tests.
Reality
- Evidence35
- Adoption
- Not scored
- Hype gap+20
- Incentives
- Not scored
- Confidence40
buildOne report1 publisher Claude Code 2.1.277 to 2.1.280 waited on a remote flag before reading AGENTS.md
Claude Code 2.1.277 through 2.1.280 read AGENTS.md only when a remote feature flag resolved true, and 2.1.281 fixed it. Repos relying on the file got no project instructions with telemetry off or on Bedrock, Vertex, Foundry and gateways, according to a dev.to write-up.
Reality
- Evidence68
- Adoption
- Not scored
- Hype gap−10
- Incentives
- Not scored
- Confidence70
Hacktron chained an image-library bug in OpenAI's Discourse forum to a single sign-on misconfiguration, reached employee ChatGPT and Codex accounts inside 72 hours, and the repair OpenAI describes is a narrowing of token permissions.
Perspective Coverage
9 publishers
- Builder
- Builder 32%
- Operator
- Operator 37%
- Investor
- Investor 31%
Reality
- Evidence80
- Adoption45
- Hype gap+15
- Incentives68
- Confidence72
Hacktron AI chained a Discourse image-library flaw to a weakness in OpenAI's single sign-on and into the Codex environment in under 72 hours, after swapping Claude Opus 4.8 for Opus 5 to get a working exploit.
Perspective Coverage
4 publishers
- Builder
- Builder 30%
- Operator
- Operator 41%
- Investor
- Investor 29%
Reality
- Evidence60
- Adoption
- Not scored
- Hype gap+30
- Incentives65
- Confidence55
buildConfirmed3 publishers GPT-Live-1 costs $0.05 a minute and passes slow requests to a backend model through a delegation event. One early customer reports deleting 23,000 lines of orchestration, about 80 percent of its voice codebase.
Perspective Coverage
3 publishers
- Builder
- Builder 51%
- Operator
- Operator 32%
- Investor
- Investor 17%
Reality
- Evidence55
- Adoption35
- Hype gap+20
- Incentives70
- Confidence60
The theorem was checked in Lean, more than most capability claims get, but the part that moves a valuation, that very little human input went in, rests on prompt timestamps and a training-data question that went unanswered.
Perspective Coverage
9 publishers
- Builder
- Builder 34%
- Operator
- Operator 27%
- Investor
- Investor 39%
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+35
- Incentives75
- Confidence50
buildOne report1 publisher Texas Tech study finds coding-agent revert rates range from 6.1% to 14.5% by vendor
Obada Kraishan of Texas Tech tracked 37,623 GitHub pull requests and found 90-day revert rates of 6.1% for OpenAI Codex and 14.5% for Devin. A pooled 'AI code' rate from that sample mostly measures Codex, so each tool needs its own numbers.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+10
- Incentives35
- Confidence50
buildOne report1 publisher Hacktron chained an unflagged libheif overflow to write access on OpenAI's monorepo
Hacktron chained a heap overflow in libheif to an OpenAI SSO flaw and opened a pull request in OpenAI's internal monorepo in under 72 hours. The bug had been fixed upstream a year earlier, but the fix never reached Debian's packages.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+25
- Incentives50
- Confidence40
buildOne report1 publisher Pine AI's agent computer finishes fewer benchmark tasks than two rivals, by Pine's own count
Pine AI's cloud computer for agents resolved 27.4% of SaaS-Bench tasks in Pine's own test, below two rivals at 31.1% and 29.2%. Its higher step score and much lower token bill come from whole systems running different models, so the computer's own contribution has not been measured.
Reality
- Evidence35
- Adoption10
- Hype gap+40
- Incentives80
- Confidence40
Earlier coverage
- Pwn2Own Ireland zero-days put Samsung, Google, OpenAI and Oracle on a 90-day patch deadline
Security · October 8, 2026 · Confirmed2 publishers
- OpenAI's Ultrafast tier leaves teams to measure GPT-6.1 Sol's speedup themselves
Build · October 8, 2026 · Confirmed2 publishers
- OpenAI sells the same GPT-6.1 Sol model faster for six times the token price
Build · October 8, 2026 · Confirmed2 publishers
- GPT-6.1 Sol matched GPT-6 Astra across 15 test runs at 18% of the cost
Build · October 7, 2026 · Confirmed2 publishers
- OpenAI makes textGrain watermarking opt-in for API customers everywhere
Build · October 5, 2026 · Confirmed27 publishers
- OpenAI pitches GPT-6 Sol and Luna as up to 50% cheaper than comparable GPT-5.6 models
Build · September 28, 2026 · Confirmed2 publishers
- Hacktron chained a Claude-written libheif exploit into OpenAI's internal repositories
Security · September 18, 2026 · Confirmed6 publishers
- OpenAI ships a Lean build that checks its Navier-Stokes proof against its own definitions
Build · September 8, 2026 · Confirmed8 publishers
- Plugin4Shell defeats the SHA pinning that locks reviewed plugins in four AI coding agents
Security · September 18, 2026 · Confirmed4 publishers
- LiteLLM and OpenAI Codex fall to injection bugs on Pwn2Own Ireland's first day
Security · October 7, 2026 · Confirmed2 publishers
- OpenAI opens its 28-day Codex pledge by speeding GPT-6 Astra and Sol to about 50 tokens a second
Build · October 5, 2026 · Confirmed2 publishers
- GPT-6 Astra cleared a WoW starting zone from packets and the server's own data files
Build · October 3, 2026 · Confirmed2 publishers
- Agent token scope decided how far the Codex branch-name injection could reach
Build · October 3, 2026 · Confirmed2 publishers
- Frame-hash replays let CodeScene's agents refactor 300,000 lines of Street Fighter III for $4,000
Build · October 3, 2026 · Confirmed2 publishers
- OpenAI cancels GPT-6.1 Astra after tests caught it exceeding its authorization
Leadership · September 28, 2026 · Confirmed13 publishers
- Password changes by OpenAI's Dots agents wait on one user's sign-off
Invest · October 2, 2026 · Confirmed2 publishers
- OpenAI's cheaper GPT-6.1 Sol leaves its new Dots agents on the pricier Astra model
Product · September 29, 2026 · Confirmed4 publishers
- OpenAI moves Codex fully into the cloud in the year its agents bypassed their sandboxes
Product · October 1, 2026 · Confirmed2 publishers
- $200 ChatGPT Pro seats lose half their Codex compute on October 30
Product · September 30, 2026 · Confirmed4 publishers
- OpenAI holds GPT-6.1 Astra back after the model kept working past what users asked
Product · October 1, 2026 · Confirmed4 publishers