The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
A Backup App Breach, Open Models Go Big, and Crypto Preps for Quantum
1Day Agency Tech Brief · October 18, 2026
A backup tool turned into a side door into Microsoft 365. Two labs shipped big open models built for coding agents. Meta rewrote its terms after a scraping scare, and crypto firms started work on quantum-safe standards. Here's what mattered in the last 48 hours.
—
1. A backup app's keys opened 162+ Microsoft 365 tenants
What happened: Attackers got into application identities inside ConnectWise Cloud Backup (formerly SkyKick) and signed in to customers' Microsoft 365 tenants as the backup app itself. Huntress counted at least 162 compromised tenants, and Microsoft confirmed unauthorized sign-in attempts starting Oct. 15. ConnectWise shut off app access and says it has no sign that backup data was taken.
Why it matters: MFA didn't help, because no person ever logged in. The app already held permission to read mail, OneDrive, SharePoint and Teams. If you run a tenant, review your third-party app permissions today.
2. DeepSeek ships a faster Flash model and opens its search engine to agents
What happened: DeepSeek upgraded its Flash model to V4.1-Flash-1017, with MIT-licensed weights and the same API name and price. It reports 77.0 on DeepSWE v1.1, up from 74.2. It also launched standalone Search and Fetch APIs at $0.75 per 1,000 successful calls, free through DeepSeek Harness, Claude Code or Codex until Nov. 1.
Why it matters: Web search for agents used to mean a separate vendor and a separate bill. Now it's a cheap add-on from the model provider.
Sources: DeepSeek API docs (https://api-docs.deepseek.com/) · Ahead of the Bot (https://aheadofthebot.com/news/2026-10-17-deepseek-v4-1-flash-1017-search-api/)
3. Reflection's Beam: a 501B open model under Apache 2.0
What happened: New US lab Reflection released Beam on Hugging Face: a mixture-of-experts model with 501B parameters (23B active per token), a 1M-token context and a focus on coding and agents. Reflection reports 81.6% on SWE-bench Verified, with day-one support in transformers, vLLM and SGLang.
Why it matters: Apache 2.0 means companies can fine-tune it and run it on their own servers. The catch is hardware: Hugging Face suggests planning for four B200 GPUs.
Sources: Hugging Face blog (https://huggingface.co/blog/reflection-beam) · Model page (https://huggingface.co/reflection/Beam-1.0-501B-A23B)
4. Meta pulls an AI scraping clause after publisher backlash
What happened: Meta's updated commercial terms said it could use content from business accounts' domains to "train" its products. Legal experts quoted by Nieman Lab called it a remarkable provision. Meta removed the language, saying the terms only cover its ads and business products and don't change how its crawlers work. Some experts say the new wording is still broad.
Why it matters: AI training rights are now decided in terms of service. Read them before you connect your business pages to any platform.
Sources: TechCrunch (https://techcrunch.com/2026/10/17/meta-removes-language-that-raised-scraping-and-ai-training-concerns/) · Nieman Lab (https://www.niemanlab.org/2026/10/legal-experts-warn-metas-new-terms-still-grant-broad-scraping-permissions-despite-language-changes/)
5. Nine crypto groups form a Quantum Alliance
What happened: Quantus, Postquant Labs, Midnight Foundation, Keystone, Abelian, Hacken and others launched an open standards body, modeled on the IETF, to define and test what "quantum-ready" means for chains, wallets and bridges. It counts about 6.04 million BTC (roughly 30% of supply) with exposed public keys.
Why it matters: No quantum computer can break Bitcoin today, but migrations take years. A shared test beats marketing claims of "quantum-resistant."
Sources: Decrypt (https://decrypt.co/381236/crypto-firms-launch-quantum-alliance-as-6-million-bitcoin-sit-exposed) · GlobeNewswire (https://www.globenewswire.com/news-release/2026/10/16/3382532/0/en/quantum-alliance-forms-to-define-and-verify-post-quantum-readiness-for-the-blockchain-industry.html)
—
𝘖𝘶𝘳 𝘵𝘢𝘬𝘦 — 1𝘋𝘢𝘺 𝘈𝘨𝘦𝘯𝘤𝘺
𝘉𝘶𝘪𝘭𝘥𝘪𝘯𝘨 𝘨𝘰𝘵 𝘤𝘩𝘦𝘢𝘱𝘦𝘳 𝘵𝘩𝘪𝘴 𝘸𝘦𝘦𝘬. 𝘛𝘳𝘶𝘴𝘵 𝘥𝘪𝘥𝘯'𝘵.
𝘈 𝘧𝘳𝘰𝘯𝘵𝘪𝘦𝘳-𝘴𝘤𝘢𝘭𝘦 𝘰𝘱𝘦𝘯 𝘮𝘰𝘥𝘦𝘭 𝘪𝘴 𝘰𝘯𝘦 𝘥𝘰𝘸𝘯𝘭𝘰𝘢𝘥 𝘢𝘸𝘢𝘺, 𝘢𝘯𝘥 𝘢𝘨𝘦𝘯𝘵𝘴 𝘤𝘢𝘯 𝘴𝘦𝘢𝘳𝘤𝘩 𝘵𝘩𝘦 𝘸𝘦𝘣 𝘧𝘰𝘳 𝘧𝘳𝘢𝘤𝘵𝘪𝘰𝘯𝘴 𝘰𝘧 𝘢 𝘤𝘦𝘯𝘵. 𝘛𝘩𝘦 𝘩𝘢𝘳𝘥 𝘱𝘢𝘳𝘵 𝘯𝘰𝘸 𝘪𝘴 𝘸𝘩𝘢𝘵 𝘺𝘰𝘶 𝘭𝘦𝘵 𝘵𝘩𝘦𝘮 𝘵𝘰𝘶𝘤𝘩. 𝘛𝘩𝘦 𝘊𝘰𝘯𝘯𝘦𝘤𝘵𝘞𝘪𝘴𝘦 𝘣𝘳𝘦𝘢𝘤𝘩 𝘤𝘢𝘮𝘦 𝘧𝘳𝘰𝘮 𝘱𝘦𝘳𝘮𝘪𝘴𝘴𝘪𝘰𝘯𝘴 𝘨𝘳𝘢𝘯𝘵𝘦𝘥 𝘺𝘦𝘢𝘳𝘴 𝘢𝘨𝘰 𝘢𝘯𝘥 𝘯𝘦𝘷𝘦𝘳 𝘳𝘦𝘷𝘪𝘦𝘸𝘦𝘥. 𝘔𝘦𝘵𝘢'𝘴 𝘤𝘭𝘢𝘶𝘴𝘦 𝘤𝘢𝘮𝘦 𝘧𝘳𝘰𝘮 𝘵𝘦𝘳𝘮𝘴 𝘯𝘰𝘣𝘰𝘥𝘺 𝘳𝘦𝘢𝘥𝘴. 𝘉𝘪𝘵𝘤𝘰𝘪𝘯'𝘴 𝘲𝘶𝘢𝘯𝘵𝘶𝘮 𝘳𝘪𝘴𝘬 𝘴𝘪𝘵𝘴 𝘪𝘯 𝘬𝘦𝘺𝘴 𝘵𝘩𝘢𝘵 𝘸𝘦𝘳𝘦 𝘧𝘪𝘯𝘦 𝘸𝘩𝘦𝘯 𝘵𝘩𝘦𝘺 𝘸𝘦𝘳𝘦 𝘮𝘢𝘥𝘦.
𝘛𝘩𝘢𝘵'𝘴 𝘩𝘰𝘸 𝘸𝘦 𝘣𝘶𝘪𝘭𝘥 𝘠𝘷𝘪𝘰𝘯: 𝘌𝘙𝘗, 𝘊𝘙𝘔 𝘢𝘯𝘥 𝘶𝘴𝘦𝘳 𝘮𝘰𝘯𝘪𝘵𝘰𝘳𝘪𝘯𝘨 𝘸𝘪𝘵𝘩 𝘢𝘤𝘤𝘦𝘴𝘴 𝘤𝘰𝘯𝘵𝘳𝘰𝘭𝘴, 𝘴𝘰 𝘦𝘷𝘦𝘳𝘺 𝘱𝘦𝘳𝘮𝘪𝘴𝘴𝘪𝘰𝘯 𝘩𝘢𝘴 𝘢𝘯 𝘰𝘸𝘯𝘦𝘳 𝘢𝘯𝘥 𝘢𝘯 𝘢𝘶𝘥𝘪𝘵 𝘵𝘳𝘢𝘪𝘭. 𝘍𝘰𝘳 𝘈𝘐 𝘢𝘯𝘥 𝘞𝘦𝘣3 𝘱𝘳𝘰𝘫𝘦𝘤𝘵𝘴, 𝘵𝘩𝘦 𝘴𝘢𝘮𝘦 𝘳𝘶𝘭𝘦 𝘢𝘱𝘱𝘭𝘪𝘦𝘴: 𝘴𝘤𝘰𝘱𝘦𝘥 𝘬𝘦𝘺𝘴, 𝘭𝘦𝘢𝘴𝘵 𝘱𝘳𝘪𝘷𝘪𝘭𝘦𝘨𝘦, 𝘢𝘯𝘥 𝘢 𝘴𝘤𝘩𝘦𝘥𝘶𝘭𝘦𝘥 𝘳𝘦𝘷𝘪𝘦𝘸 𝘰𝘧 𝘸𝘩𝘰 𝘤𝘢𝘯 𝘥𝘰 𝘸𝘩𝘢𝘵. 𝘚𝘩𝘪𝘱 𝘧𝘢𝘴𝘵, 𝘣𝘶𝘵 𝘳𝘦-𝘤𝘩𝘦𝘤𝘬 𝘺𝘰𝘶𝘳 𝘥𝘰𝘰𝘳𝘴.
Want this applied to your business? Message us here on Contra.
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
AI lead triage in n8n: webhook → validate → LLM hot/warm/cold → HubSpot upsert → Slack alert, plus an error branch so nothing fails silently. Works with any OpenAI-compatible endpoint. You bring your own API keys.
n8n AI Lead Capture to CRM Template · $39.99 — search Logan Bishop on Contra to grab it, or check the Products tab on my profile.
#n8n #aiautomation #leadgeneration #hubspot #nocode #workflowautomation
The error branch preventing silent failures is a crucial detail. Most basic automation setups skip that, which leads to mysterious lead loss that is difficult to trace later.
I built an agent that scopes app ideas. Fed a typical studio owner's wishlist, the first thing it cut was the AI. Once people can book on their own, the DMs stop coming.
Most agencies think AI SEO is just about adding Schema markup.
They are missing the main part: How AI actually connects brands to topics.
Here is how ChatGPT and Perplexity decide which business to recommend:
Brand Topic Connection: AI builds a mental map. If your brand isn't constantly mentioned next to your core service online, AI won't trust you.
Entity Sources: AI checks structured sources like Wikidata, Crunchbase, and niche directories to verify if your business is real and authoritative.
Co-mentions: If your brand is mentioned alongside top industry leaders across blogs and forums, AI treats you as a top choice.
Traditional SEO helps you rank on Google.
Entity optimization helps you get recommended by AI.
Want my AI Brand Visibility Checklist to see if AI trusts your brand?
Comment "AI" below and I’ll DM it to you!
Agree on the topic connection part. We saw the same with B2B lists: a brand gets named when third party pages describe it consistently, not when the markup is perfect.