<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Steve Antoine</title>
    <description>The latest articles on DEV Community by Steve Antoine (@epmosgateless).</description>
    <link>https://dev.to/epmosgateless</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4252349%2F95e6d7d4-44f7-43f4-a277-3dca62598b68.jpg</url>
      <title>DEV Community: Steve Antoine</title>
      <link>https://dev.to/epmosgateless</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/epmosgateless"/>
    <language>en</language>
    <item>
      <title>EPMOS Daily World-Scan — Wednesday, November 18, 2026</title>
      <dc:creator>Steve Antoine</dc:creator>
      <pubDate>Wed, 18 Nov 2026 17:34:14 +0000</pubDate>
      <link>https://dev.to/epmosgateless/epmos-daily-world-scan-wednesday-november-18-2026-3o1k</link>
      <guid>https://dev.to/epmosgateless/epmos-daily-world-scan-wednesday-november-18-2026-3o1k</guid>
      <description>&lt;blockquote&gt;
&lt;p&gt;Publication note: one or more items from today's sweep were withheld from this publication because their source could not be verified against the labeled source today. They will be published when the primary source is confirmed.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h1&gt;
  
  
  EPMOS Daily World-Scan — Wednesday, November 18, 2026
&lt;/h1&gt;

&lt;h2&gt;
  
  
  Verified developments
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Nov 18 — This smell startup says its AI nose is as good as a human's — MIT Technology Review.&lt;/li&gt;
&lt;li&gt;Nov 17 — OpenAI announced it introduced GPT‑6.2 Astra and Sol and published the system card the same day — OpenAI News.&lt;/li&gt;
&lt;li&gt;Nov 17 — OpenAI announced the launch of GPT‑6.2 Astra and Sol — OpenAI News.&lt;/li&gt;
&lt;li&gt;Nov 17 — OpenAI says it is working to strengthen NATO and allied cyber defenses with AI — OpenAI News.&lt;/li&gt;
&lt;li&gt;Nov 17 — Nvidia's startup stakes grew to $65 billion, but Jensen Huang's exit raises uncertainty for future deals — TechCrunch.&lt;/li&gt;
&lt;li&gt;Nov 17 — Nvidia says its revenue will continue to rise for the near term — TechCrunch.&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Held on Radar (Pending Primary Source Verification)
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;[RADAR ONLY] Hacker News discussion shows traction on Cerebras Disaggregated Inference; primary source (the study) still needed.
&lt;/li&gt;
&lt;li&gt;[RADAR ONLY] Hacker News discussion shows traction on Google’s AI answers reducing search traffic by 40%; primary source (the transcript) still needed.
&lt;/li&gt;
&lt;li&gt;[RADAR ONLY] Hacker News discussion shows traction on Bank CEOs, NSA telling Congress that AI‑driven patches are behind outages; primary source (the wire confirmation) still needed.
&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Projections — labeled, not facts
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;[PROJECTION] Basis: According to OpenAI News, OpenAI announced GPT‑6.2 Astra and Sol on 17 November 2026. Would change if the rollout experiences significant delays or regulatory constraints.
&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  EPMOS Perspective — what this means for our own roadmap and pitch
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;According to MIT Technology Review, the emergence of an AI nose comparable to human olfaction suggests new multimodal sensor integration opportunities for our platform.
&lt;/li&gt;
&lt;li&gt;According to OpenAI News, the introduction of GPT‑6.2 Astra and Sol indicates a shift toward more specialized large‑model families, prompting us to consider modular extensions in our offering.
&lt;/li&gt;
&lt;li&gt;According to TechCrunch, Nvidia’s revenue optimism signals continued market appetite for AI hardware, reinforcing the case for aligning our compute‑as‑a‑service strategy with GPU‑centric workloads.
&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  Method &amp;amp; limits
&lt;/h2&gt;

&lt;p&gt;Sweep facts for today:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;TechCrunch — AI: ok (8 recent articles)&lt;/li&gt;
&lt;li&gt;MIT Technology Review — Artificial intelligence: ok (5 recent items)&lt;/li&gt;
&lt;li&gt;Reuters — Artificial Intelligence: unfetchable (HTTP 401 via curl; governed Playwright fetch returned an empty page on 2026-11-17)&lt;/li&gt;
&lt;li&gt;arXiv — cs.AI recent: ok (5 recent papers)&lt;/li&gt;
&lt;li&gt;OpenAI News: ok (8 recent items)&lt;/li&gt;
&lt;li&gt;Anthropic Newsroom: ok (0 recent articles)&lt;/li&gt;
&lt;li&gt;Google AI Blog: ok (2 recent items)&lt;/li&gt;
&lt;li&gt;Hacker News: ok (8 recent stories)&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;Open question: Details beyond the headline were not verified in this sweep?&lt;/p&gt;

&lt;h2&gt;
  
  
  Sources
&lt;/h2&gt;

&lt;p&gt;MIT Technology Review — &lt;a href="https://www.technologyreview.com/2026/11/18/1148816/smell-startup-osmo-ai-nose-good-as-humans/" rel="noopener noreferrer"&gt;https://www.technologyreview.com/2026/11/18/1148816/smell-startup-osmo-ai-nose-good-as-humans/&lt;/a&gt;&lt;br&gt;&lt;br&gt;
OpenAI News — &lt;a href="https://openai.com/news/" rel="noopener noreferrer"&gt;https://openai.com/news/&lt;/a&gt;&lt;br&gt;&lt;br&gt;
OpenAI News — &lt;a href="https://openai.com/index/introducing-gpt-6-2-astra-sol" rel="noopener noreferrer"&gt;https://openai.com/index/introducing-gpt-6-2-astra-sol&lt;/a&gt;&lt;br&gt;&lt;br&gt;
OpenAI News — &lt;a href="https://openai.com/index/strengthening-nato-allied-cyber-defences-ai" rel="noopener noreferrer"&gt;https://openai.com/index/strengthening-nato-allied-cyber-defences-ai&lt;/a&gt;&lt;br&gt;&lt;br&gt;
TechCrunch — &lt;a href="https://techcrunch.com/2026/11/17/nvidias-startup-stakes-ballooned-to-65b-but-huangs-exit-puts-future-deals-in-doubt/" rel="noopener noreferrer"&gt;https://techcrunch.com/2026/11/17/nvidias-startup-stakes-ballooned-to-65b-but-huangs-exit-puts-future-deals-in-doubt/&lt;/a&gt;&lt;br&gt;&lt;br&gt;
TechCrunch — &lt;a href="https://techcrunch.com/2026/11/17/nvidia-promises-that-revenue-will-keep-soaring-for-now/" rel="noopener noreferrer"&gt;https://techcrunch.com/2026/11/17/nvidia-promises-that-revenue-will-keep-soaring-for-now/&lt;/a&gt;&lt;br&gt;&lt;br&gt;
Hacker News — &lt;a href="https://news.ycombinator.com/item?id=50540472" rel="noopener noreferrer"&gt;https://news.ycombinator.com/item?id=50540472&lt;/a&gt;&lt;br&gt;&lt;br&gt;
Hacker News — &lt;a href="https://news.ycombinator.com/item?id=50540178" rel="noopener noreferrer"&gt;https://news.ycombinator.com/item?id=50540178&lt;/a&gt;&lt;br&gt;&lt;br&gt;
Hacker News — &lt;a href="https://news.ycombinator.com/item?id=50540836" rel="noopener noreferrer"&gt;https://news.ycombinator.com/item?id=50540836&lt;/a&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>news</category>
    </item>
    <item>
      <title>Can Your AI Survive 10,000 Adversarial Tests? The EPMOS Harness Is Open</title>
      <dc:creator>Steve Antoine</dc:creator>
      <pubDate>Wed, 18 Nov 2026 17:13:10 +0000</pubDate>
      <link>https://dev.to/epmosgateless/can-your-ai-survive-10000-adversarial-tests-the-epmos-harness-is-open-5c7l</link>
      <guid>https://dev.to/epmosgateless/can-your-ai-survive-10000-adversarial-tests-the-epmos-harness-is-open-5c7l</guid>
      <description>&lt;h1&gt;
  
  
  Can Your AI Survive 10,000 Adversarial Tests?
&lt;/h1&gt;

&lt;p&gt;Most AI systems are shipped on the strength of a demo. This challenge asks a different question: what happens when the system is tested, case by case, against inputs designed to find where it breaks?&lt;/p&gt;

&lt;p&gt;EPMOS has opened the EPMOS Harness. You do not build something new to enter. You bring an AI system you already built, and you find out how it holds up. You participate right through EPMOS — no Kaggle account is needed to enter.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Also listed on Kaggle:&lt;/strong&gt; &lt;a href="https://www.kaggle.com/competitions/can-your-ai-survive-10-000-adversarial-tests" rel="noopener noreferrer"&gt;https://www.kaggle.com/competitions/can-your-ai-survive-10-000-adversarial-tests&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;GitHub:&lt;/strong&gt; &lt;a href="https://github.com/EpmosGateless/epmos-harness-hackathon" rel="noopener noreferrer"&gt;https://github.com/EpmosGateless/epmos-harness-hackathon&lt;/a&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Harness:&lt;/strong&gt; &lt;a href="https://epmos.site/diagnosed" rel="noopener noreferrer"&gt;https://epmos.site/diagnosed&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;h2&gt;
  
  
  The case set
&lt;/h2&gt;

&lt;p&gt;The set is frozen: &lt;code&gt;epmos-ai-adversarial/v1&lt;/code&gt;, seed &lt;code&gt;20261116&lt;/code&gt;. 10,000 cases across 8 families:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Family&lt;/th&gt;
&lt;th&gt;What it probes&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;Input handling&lt;/td&gt;
&lt;td&gt;Ordinary and edge-case inputs&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Boundary enforcement&lt;/td&gt;
&lt;td&gt;Requests that cross a stated boundary&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Malformed input&lt;/td&gt;
&lt;td&gt;Broken, truncated, and unexpected shapes&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Link/fetch&lt;/td&gt;
&lt;td&gt;URLs and fetched content&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Hold/gate&lt;/td&gt;
&lt;td&gt;Conditions that should make the system stop and hold&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Oversized body&lt;/td&gt;
&lt;td&gt;Payloads larger than the system should accept&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Anomaly detection&lt;/td&gt;
&lt;td&gt;Inputs that do not fit the expected pattern&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;Repair verification&lt;/td&gt;
&lt;td&gt;Whether a repaired system actually holds&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;p&gt;The free 1,000 is the proportional subset (240 / 160 / 120 / 80 / 120 / 40 / 40 / 200). Every participant faces the same cases.&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it right now — free
&lt;/h2&gt;

&lt;p&gt;You expose your system at a public endpoint. The harness POSTs each case as JSON and records the response. Your prompts, logic, and weights never leave your boundary.&lt;/p&gt;

&lt;p&gt;Clone the repo and run one command:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;git clone https://github.com/EpmosGateless/epmos-harness-hackathon.git
&lt;span class="nb"&gt;cd&lt;/span&gt; epmos-harness-hackathon
./try-it.sh https://your-ai.example.com/chat
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;&lt;code&gt;try-it.sh&lt;/code&gt; is pure Python 3, standard library only. It submits the free 1,000-case run, polls the live harness, and prints your receipt. No sign-up, no key, no payment. Against a fast endpoint the free run takes about 3 seconds.&lt;/p&gt;

&lt;p&gt;Prefer curl? The API is:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl -s -X POST https://epmos.site/api/ai-run \
  -H &lt;span class="s1"&gt;'content-type: application/json'&lt;/span&gt; \
  -d &lt;span class="s1"&gt;'{"system_name":"my-ai","system_version":"1.0.0","endpoint_url":"https://your-ai.example.com/chat","tier":"free-1k"}'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;Then poll for the receipt:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight shell"&gt;&lt;code&gt;curl -s https://epmos.site/api/ai-run/&lt;span class="o"&gt;&amp;lt;&lt;/span&gt;run_id&lt;span class="o"&gt;&amp;gt;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The endpoint contract: the harness POSTs &lt;code&gt;{"case_id", "family", "subfamily", "input"}&lt;/code&gt; to your URL. It must be a public &lt;code&gt;http(s)&lt;/code&gt; URL — private, loopback, link-local, and reserved addresses are refused, and the URL must not carry credentials. 15-second per-case timeout.&lt;/p&gt;

&lt;h2&gt;
  
  
  What you get
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Your first 1,000 cases free.&lt;/strong&gt; A mini smoke run: find the kinks, fix them, run again. A re-run after a repair carries no penalty.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;The 10,000-case qualification for $1,000.&lt;/strong&gt; That buys the run, not the badge. If your system passes the full qualification, you earn an EPMOS badge for that specific system and version. The badge is awarded by the EPMOS qualification process, not by Kaggle. A new version or a material change must earn it again.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;A receipt for every run.&lt;/strong&gt; Hash-chained, recording the exact case count that ran, how many passed and how many failed. Failures are preserved with their reason and repair. Nothing is erased to make a result look clean.&lt;/p&gt;

&lt;h2&gt;
  
  
  Provenance
&lt;/h2&gt;

&lt;p&gt;This is the part that matters most to us. The harness is not a claim; it has run:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;10,000-case AWS smoke run, November 14, 2026: 10,000 pass, 0 fail.&lt;/li&gt;
&lt;li&gt;1,000,000-case AWS qualification run, November 14, 2026: 1,000,000 pass, 0 fail.&lt;/li&gt;
&lt;li&gt;Harness Repair Library v1.0.5: 202 failures, 167 repairs, 103 regression cases, validation 41/41 pass.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The competition runs to November 16, 2027. Enter right through EPMOS — run ./try-it.sh or POST &lt;a href="https://epmos.site/api/ai-run" rel="noopener noreferrer"&gt;https://epmos.site/api/ai-run&lt;/a&gt; — try it free on GitHub, and bring the system you already have.&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Disclosure: this is the EPMOS competition. The code for the instant try is in the repo; the case set and the receipt format are documented there too.&lt;/em&gt;&lt;/p&gt;

</description>
      <category>kaggle</category>
      <category>ai</category>
      <category>testing</category>
      <category>opensource</category>
    </item>
  </channel>
</rss>
