buildOne report1 publisher Forwarding support mail under its own DKIM key let every spam pitch into CogniPrep's inbox
CogniPrep's support forwarder sent inbound mail onward under its own DKIM key, so 100% of mail to its catch-all address, junk included, reached the inbox. Teams relaying mail under their own domain blind the receiving filter the same way and must score it themselves.
Reality
- Evidence45
- Adoption5
- Hype gap0
- Incentives
- Not scored
- Confidence50
buildOne report1 publisher One unthrottled signup endpoint gave a bot six API keys in six hours
A developer's signup route issued six API keys overnight to a bot using dot-mutated Gmail addresses. Stripped of dots they were six strangers' inboxes, so canonicalizing emails would not have stopped it.
Reality
- Evidence50
- Adoption
- Not scored
- Hype gap+5
- Incentives
- Not scored
- Confidence55
The code containers could not reach the internet or each other, but every one of them reached the same package service, and its metadata was not scoped by account, so a planted prompt turned one user's assistant into a stranger's Gmail reader.
Perspective Coverage
3 publishers
- Builder
- Builder 37%
- Operator
- Operator 53%
- Investor
- Investor 10%
Reality
- Evidence64
- Adoption
- Not scored
- Hype gap+15
- Incentives55
- Confidence68
Meta's Muse stopped at the payment screen on a $7 espresso part
The Verge gave Meta's new agent a backlog of postponed chores. Muse asked for Gmail access, found the part and drafted the email, and both of the final steps, send and pay, stayed with the human.
Reality
- Evidence45
- Adoption15
- Hype gap+20
- Incentives25
- Confidence55
buildOne report1 publisher Calendar clients ignore an update whose SEQUENCE is not higher than the one they already hold
A Node.js walkthrough traces duplicate events and cancellations that never land to three fields in RFC 5546. Getting them right costs a UID and a sequence counter stored per event, plus one regex that reads METHOD back out of the file.
Reality
- Evidence57
- Adoption
- Not scored
- Hype gap+14
- Incentives66
- Confidence54
Googlebook OS opens Gmail and Calendar as web wrappers from the Play Store
9to5Google's hands-on finds an Android shell down to the Pixel launcher and the desktop widgets, while the first-party apps a new user reaches for first are Trusted Web Activities the Workspace team picked over its own Android build.
Reality
- Evidence58
- Adoption15
- Hype gap+12
- Incentives60
- Confidence55
Google shipped Gmail on Googlebook OS as a Play Store web wrapper
On hands-on units, Gmail and Calendar open as Trusted Web Activities distributed through the Play Store, with Gemini Nano doing the on-device work. Google says each app team picks its own format.
Reality
- Evidence55
- Adoption20
- Hype gap+10
- Incentives60
- Confidence55
Grok Bot ties write access to Gmail and Salesforce to a personal paid seat
Engadget says SpaceXAI's agent signs into Gmail, LinkedIn and Salesforce and acts for the user, with eligibility running through six individual paid tiers. The only brake described is a human-approval setting.
Reality
- Evidence30
- Adoption15
- Hype gap+20
- Incentives55
- Confidence35
buildOne report1 publisher Counting the full SPF include tree nightly catches the vendor include that breaks your record
The DNS and header half of deliverability is a pure function of your own configuration, so a scheduled job can assert it; inbox placement is a measurement you put on a dashboard with a date on it.
Reality
- Evidence52
- Adoption
- Not scored
- Hype gap+10
- Incentives55
- Confidence55
Itamar Gilad draws an AI danger zone where output outruns the reviewer's judgement
An essay by the former Google product manager argues that LLMs lift what a person can produce faster than what they can evaluate. Every curve in it is labelled speculative, and it reports no measurement of the gap.
Publishers:itamargilad.com
Reality
- Evidence32
- Adoption
- Not scored
- Hype gap+15
- Incentives30
- Confidence55
buildOne report1 publisher OpenAI sets October 14, 2026 as the last day GPT-5.5 runs a scheduled task
The retirement covers ChatGPT, ChatGPT Work and Codex on all plans. OpenAI names gpt-5.6-sol as the replacement for Codex with ChatGPT sign-in. The API is exempt, and the inventory of affected tasks lives in one screen.
Reality
- Evidence72
- Adoption
- Not scored
- Hype gap−10
- Incentives60
- Confidence68
buildOne report1 publisher Repository-item properties in one internal Artifactory carried data between two ChatGPT accounts
Check Point's Alexey Bukhteyev told Lets Data Science that the test containers could not reach each other directly while both could still write metadata to the same package service, using credentials already present inside them.
Reality
- Evidence52
- Adoption15
- Hype gap−8
- Incentives58
- Confidence55
buildOne report1 publisher A live Gmail connector cost four of five assistants their inbox-triage pass
RuntimeWire ran five assistants through the same nine tasks on September 16th with a single GPT-6 judge. All five refused the fake invoice and the lookalike payee; only Claude Cowork read the inbox the pack supplied.
Reality
- Evidence42
- Adoption24
- Hype gap+12
- Incentives55
- Confidence46
buildOne report1 publisher Past thirty tools, MCP servers' tool descriptions start to blur together
A scan of 4,951 public MCP servers and 87,146 tools finds description overlap climbing with tool count, until nearly one tool in three on the biggest servers has no word its siblings do not also use. The study measured text; it did not measure model choices.
Reality
- Evidence66
- Adoption30
- Hype gap+15
- Incentives70
- Confidence58
Gemini takes over as the default voice layer for Android Auto users everywhere
Engadget reports the swap from Google's regular assistant is rolling out to Android Auto users everywhere. Drivers who never opened the setting now get a handler that infers intent instead of matching a memorised phrase.
Reality
- Evidence32
- Adoption30
- Hype gap+24
- Incentives
- Not scored
- Confidence38
buildOne report1 publisher Check Point pulled Gmail data across ChatGPT accounts through a shared package-metadata store
OpenAI isolated code-interpreter containers per user account, then let all of them read and write the same Artifactory-backed package metadata so they could install software. The metadata carried an instruction into someone else's session.
Reality
- Evidence25
- Adoption30
- Hype gap+15
- Incentives60
- Confidence32
buildOne report1 publisher Port 25 decides which email check you can actually run
An MX lookup is one DNS query proving a domain has mail servers. Confirming a specific mailbox means opening port 25 from an IP that Gmail will answer, so MailProbe ships that check only in its self-hosted build.
Reality
- Evidence40
- Adoption
- Not scored
- Hype gap+10
- Incentives70
- Confidence55
Claude's Gmail agent turns one approval toggle into your whole outbound policy
Claude will send, reply and forward once you switch approval off, which leaves that review step doing double duty as the feature's only brake and as the sole defence anyone can name against prompt injection.
Reality
- Evidence42
- Adoption
- Not scored
- Hype gap+12
- Incentives35
- Confidence45
Spammers pick up the invisible Unicode block that AI jailbreakers found first
The Tags block was drafted for language codes, dropped, revived for flag emoji, then dropped again, and it still travels through every text field you run. That is why the answer is normalizing input rather than scanning it.
Reality
- Evidence52
- Adoption
- Not scored
- Hype gap+22
- Incentives18
- Confidence56
buildOne report1 publisher Base64 uploads past 6,000 characters decode cleanly but with wrong content inside
Two weeks of field notes from an agent with write access to a construction company's books and bank account point the reliability work at verification, because every failure in the log reported success first.
Reality
- Evidence30
- Adoption10
- Hype gap+18
- Incentives45
- Confidence40