<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:wfw="http://wellformedweb.org/CommentAPI/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	xmlns:slash="http://purl.org/rss/1.0/modules/slash/"
	>

<channel>
	<title>Trends &#8211; Firethering</title>
	<atom:link href="https://firethering.com/tech/trends/feed/" rel="self" type="application/rss+xml" />
	<link>https://firethering.com</link>
	<description>Firethering is Your Hub for AI, Open Source and Tech That Actually Matters</description>
	<lastBuildDate>Thu, 04 Jun 2026 12:02:38 +0000</lastBuildDate>
	<language>en-US</language>
	<sy:updatePeriod>
	hourly	</sy:updatePeriod>
	<sy:updateFrequency>
	1	</sy:updateFrequency>
	<generator>https://wordpress.org/?v=6.6.5</generator>

<image>
	<url>https://firethering.com/wp-content/uploads/2024/10/cropped-firethering-FTR-favicon-32x32.png</url>
	<title>Trends &#8211; Firethering</title>
	<link>https://firethering.com</link>
	<width>32</width>
	<height>32</height>
</image> 
	<item>
		<title>Google Built Gemma 4 12B Without Multimodal Encoders</title>
		<link>https://firethering.com/google-built-gemma-4-12b-without-multimodal-encoders/</link>
					<comments>https://firethering.com/google-built-gemma-4-12b-without-multimodal-encoders/#respond</comments>
		
		<dc:creator><![CDATA[Mohit Geryani]]></dc:creator>
		<pubDate>Thu, 04 Jun 2026 12:01:22 +0000</pubDate>
				<category><![CDATA[Tech]]></category>
		<category><![CDATA[AI Models]]></category>
		<category><![CDATA[Trends]]></category>
		<category><![CDATA[AI]]></category>
		<category><![CDATA[Google]]></category>
		<guid isPermaLink="false">https://firethering.com/?p=7257</guid>

					<description><![CDATA[Every multimodal model you've used has the same basic system. Text goes in one way, images go through a vision encoder first, audio goes through an audio encoder first, and then everything gets handed off to the language model in a form it can work with. The encoders are load-bearing and you don't just remove them.Google actually removed them.Gemma 4 12B takes raw image patches and raw audio waveforms and projects them directly into the same embedding space as text tokens. There is no vision encoder or audio encoder. One decoder handling everything.]]></description>
		
					<wfw:commentRss>https://firethering.com/google-built-gemma-4-12b-without-multimodal-encoders/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>MiniMax M3 Shows What Happens When AI Stops Thinking in Turns</title>
		<link>https://firethering.com/minimax-m3-open-weight-model/</link>
					<comments>https://firethering.com/minimax-m3-open-weight-model/#respond</comments>
		
		<dc:creator><![CDATA[Mohit Geryani]]></dc:creator>
		<pubDate>Tue, 02 Jun 2026 19:25:39 +0000</pubDate>
				<category><![CDATA[Tech]]></category>
		<category><![CDATA[AI Models]]></category>
		<category><![CDATA[Trends]]></category>
		<category><![CDATA[AI]]></category>
		<category><![CDATA[AI Agents]]></category>
		<guid isPermaLink="false">https://firethering.com/?p=7236</guid>

					<description><![CDATA[Most models quit around submission 30 because they stop finding improvement and exit on their own. That's what happened when MiniMax ran a CUDA kernel optimization task against a field of frontier models. Every model except two called it done within the first 30 submissions.

M3's best result came on submission 145. After 24 hours. After multiple plateaus where the numbers stopped moving and a reasonable model would have concluded there was nothing left to find.

That's the thing MiniMax released yesterday. An AI model with a 1M token context window, native multimodality, and apparently a problem with knowing when to stop.]]></description>
		
					<wfw:commentRss>https://firethering.com/minimax-m3-open-weight-model/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>Anthropic Files for an IPO. AI Is Entering Its Public Company Era.</title>
		<link>https://firethering.com/anthropic-ipo-ai-public-company-era/</link>
					<comments>https://firethering.com/anthropic-ipo-ai-public-company-era/#respond</comments>
		
		<dc:creator><![CDATA[Mohit Geryani]]></dc:creator>
		<pubDate>Mon, 01 Jun 2026 19:05:12 +0000</pubDate>
				<category><![CDATA[Tech]]></category>
		<category><![CDATA[Trends]]></category>
		<category><![CDATA[AI]]></category>
		<category><![CDATA[Anthropic]]></category>
		<guid isPermaLink="false">https://firethering.com/?p=7232</guid>

					<description><![CDATA[Anthropic has officially taken its first step toward becoming a public company.

In a brief announcement on Monday, the company said it had confidentially submitted a draft S-1 registration statement to the U.S. Securities and Exchange Commission for a proposed initial public offering. The filing doesn't reveal a share price, a fundraising target, or even a timeline. For now, it simply gives Anthropic the option to go public once the SEC review process is complete.

Just a few years ago, Anthropic was a small group of former OpenAI researchers trying to build an alternative vision for advanced AI. Today, it sits among the handful of companies shaping the industry's future and that's why this filing matters.

It's one of the world's most influential AI labs beginning the transition from a privately funded research company to a business that may eventually answer to public shareholders.

For most of the AI boom, the biggest bets were made behind closed doors. Venture firms, sovereign wealth funds, and tech giants supplied the capital while the public watched from the outside. Anthropic's filing suggests that era may be starting to change.]]></description>
		
					<wfw:commentRss>https://firethering.com/anthropic-ipo-ai-public-company-era/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>OpenAI Says Its AI Solved an 80-Year-Old Math Problem. The Proof Surprised Mathematicians.</title>
		<link>https://firethering.com/openai-ai-solves-80-year-old-math-problem/</link>
					<comments>https://firethering.com/openai-ai-solves-80-year-old-math-problem/#respond</comments>
		
		<dc:creator><![CDATA[Mohit Geryani]]></dc:creator>
		<pubDate>Mon, 01 Jun 2026 16:53:38 +0000</pubDate>
				<category><![CDATA[Tech]]></category>
		<category><![CDATA[Trends]]></category>
		<category><![CDATA[AI]]></category>
		<category><![CDATA[OpenAI]]></category>
		<guid isPermaLink="false">https://firethering.com/?p=7219</guid>

					<description><![CDATA[OpenAI says one of its internal reasoning models has solved a math problem that has been there on mathematicians' desks since 1946.

The problem, first posed by legendary mathematician Paul Erdős, looks almost absurdly simple. Given a set of points on a flat plane, how many pairs can be exactly one unit apart? People have spent nearly 80 years trying to pin down the answer.

OpenAI's model didn't just make progress on the problem. According to the company, it disproved a longstanding conjecture that many researchers believed was essentially correct.]]></description>
		
					<wfw:commentRss>https://firethering.com/openai-ai-solves-80-year-old-math-problem/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>StepFun Says Step 3.7 Flash Matches 97% of Claude Opus 4.6&#8217;s Coding Performance at One-Ninth the Cost</title>
		<link>https://firethering.com/stepfun-step-3-7-flash-agentic-coding-cost-efficiency/</link>
					<comments>https://firethering.com/stepfun-step-3-7-flash-agentic-coding-cost-efficiency/#respond</comments>
		
		<dc:creator><![CDATA[Mohit Geryani]]></dc:creator>
		<pubDate>Sat, 30 May 2026 16:58:32 +0000</pubDate>
				<category><![CDATA[Tech]]></category>
		<category><![CDATA[AI Models]]></category>
		<category><![CDATA[Trends]]></category>
		<category><![CDATA[AI]]></category>
		<category><![CDATA[AI Agents]]></category>
		<guid isPermaLink="false">https://firethering.com/?p=7179</guid>

					<description><![CDATA[$0.19 vs $1.76. That's the per-task cost of running Step 3.7 Flash with Advisor Mode enabled versus Claude Opus 4.6 on SWE-Bench Verified. The Flash model scores 76.3% to Opus 4.6's 78.7%. Two percentage points of difference. Nine times cheaper to get there.

For anyone building agentic coding workflows at scale that math changes the decision about which model actually belongs in production. Frontier performance has been getting cheaper for a while but this is a specific, benchmarked claim with a specific cost figure attached.]]></description>
		
					<wfw:commentRss>https://firethering.com/stepfun-step-3-7-flash-agentic-coding-cost-efficiency/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>Microsoft Threatened Legal Action Against a Security Researcher. The Security Community Pushed Back.</title>
		<link>https://firethering.com/microsoft-threatens-legal-action-nightmare-eclipse-security-researcher/</link>
					<comments>https://firethering.com/microsoft-threatens-legal-action-nightmare-eclipse-security-researcher/#respond</comments>
		
		<dc:creator><![CDATA[Mohit Geryani]]></dc:creator>
		<pubDate>Fri, 29 May 2026 20:32:09 +0000</pubDate>
				<category><![CDATA[Tech]]></category>
		<category><![CDATA[Trends]]></category>
		<category><![CDATA[Microsoft]]></category>
		<category><![CDATA[Security]]></category>
		<guid isPermaLink="false">https://firethering.com/?p=7166</guid>

					<description><![CDATA[Finding bugs in Microsoft products used to come with a clear social contract. You find it, you report it privately, you wait for a fix, then you publish. Microsoft gets to patch quietly. You get credit and maybe a bug bounty. Nowadays that contract seem to get complicated.

A researcher going by Nightmare Eclipse published a series of unpatched vulnerabilities in Microsoft products including Windows Defender and BitLocker, along with working exploit code, without giving Microsoft a chance to fix them first. Microsoft responded with a blog post threatening criminal referrals and invoking its Digital Crimes Unit.

The cybersecurity community, the same community Microsoft depends on to find these bugs before actual criminals do, reacted about as well as you'd expect.]]></description>
		
					<wfw:commentRss>https://firethering.com/microsoft-threatens-legal-action-nightmare-eclipse-security-researcher/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>The $500K AI Film That &#8216;Premiered at Cannes&#8217; Didn&#8217;t Actually Premiere at Cannes</title>
		<link>https://firethering.com/hell-grind-ai-film-cannes-premiere-higgsfield/</link>
					<comments>https://firethering.com/hell-grind-ai-film-cannes-premiere-higgsfield/#respond</comments>
		
		<dc:creator><![CDATA[Mohit Geryani]]></dc:creator>
		<pubDate>Fri, 29 May 2026 09:31:20 +0000</pubDate>
				<category><![CDATA[Tech]]></category>
		<category><![CDATA[Trends]]></category>
		<category><![CDATA[AI]]></category>
		<guid isPermaLink="false">https://firethering.com/?p=7149</guid>

					<description><![CDATA[Last week an AI startup called Higgsfield announced it had premiered a fully AI-generated feature film at Cannes. The Wall Street Journal covered it. The founder posted on LinkedIn that "for decades, Cannes has been the room where new cinema gets legitimized." The story spread fast.

There was one problem. Cannes said it never happened.

"We can confirm that 'Hell Grind' was not screened as part of the official Festival de Cannes program," a festival spokesperson said. The film was shown at a paid third-party screening at a local theater in the town of Cannes during the festival period. That's a meaningfully different thing and the distinction matters because the entire credibility of the announcement rested on the Cannes name.

This deserves the attention because it's a clean example of how AI hype gets manufactured and how quickly it travels before anyone checks.]]></description>
		
					<wfw:commentRss>https://firethering.com/hell-grind-ai-film-cannes-premiere-higgsfield/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>Your Car Knows More About You Than You Think. Insurance Companies Are Using That Data</title>
		<link>https://firethering.com/car-data-privacy-insurance-companies-using-data/</link>
					<comments>https://firethering.com/car-data-privacy-insurance-companies-using-data/#respond</comments>
		
		<dc:creator><![CDATA[Mohit Geryani]]></dc:creator>
		<pubDate>Fri, 29 May 2026 07:29:56 +0000</pubDate>
				<category><![CDATA[Tech]]></category>
		<category><![CDATA[Trends]]></category>
		<category><![CDATA[Cars]]></category>
		<category><![CDATA[Security]]></category>
		<guid isPermaLink="false">https://firethering.com/?p=7144</guid>

					<description><![CDATA[According to BBC reporting, there's a man who got a copy of his driving data from a company called LexisNexis. It was 130 pages long. Six months of every trip he and his wife took, logged, packaged, and sold without them knowing. Shortly after, his insurance costs jumped 21%. An insurance agent confirmed the data was a factor.

He hadn't signed anything that felt like permission. He'd just set up his car's infotainment system.

That's where we are with car privacy in 2026. Modern vehicles are collecting your location, your speed, how hard you brake, who's sitting next to you, and in some cases your weight, age, facial expressions, and driving patterns. Mozilla examined 25 car brands and found every single one failed its privacy and security standards. Cars, Mozilla concluded, were the worst product category it had ever reviewed for privacy. And most people have no idea any of this is happening.]]></description>
		
					<wfw:commentRss>https://firethering.com/car-data-privacy-insurance-companies-using-data/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>Nvidia Promised $500B for US AI. Its Next $150B Bet Is Still Taiwan.</title>
		<link>https://firethering.com/nvidia-500b-us-ai-investment-150b-taiwan/</link>
					<comments>https://firethering.com/nvidia-500b-us-ai-investment-150b-taiwan/#respond</comments>
		
		<dc:creator><![CDATA[Mohit Geryani]]></dc:creator>
		<pubDate>Thu, 28 May 2026 11:21:03 +0000</pubDate>
				<category><![CDATA[Tech]]></category>
		<category><![CDATA[Trends]]></category>
		<guid isPermaLink="false">https://firethering.com/?p=7123</guid>

					<description><![CDATA[Earlier this year Jensen Huang wrote a big check. Five hundred billion dollars committed to US data centers, announced with the kind of fanfare that makes a president happy and keeps tariff threats at bay. Trump called Huang brilliant. Export controls on some Nvidia chips got walked back. Everyone went home satisfied.

Then Huang flew to Taiwan, broke ground on a new Nvidia headquarters, and according to Reuters, announced the company would be spending $150 billion a year there. He called Taiwan the epicenter of the AI revolution. He said it's where the chips come from, where the packaging happens, where AI supercomputers get built. He said Nvidia would be worth even more in three to five years because of it.

Nobody in the Trump administration has commented on that yet. Nvidia didn't respond to questions about the tension between the two announcements. But the tension is there whether anyone acknowledges it or not, and it tells you something about where AI infrastructure actually lives versus where politicians want it to live.]]></description>
		
					<wfw:commentRss>https://firethering.com/nvidia-500b-us-ai-investment-150b-taiwan/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>A Critical Bug in a 325M-Download Package Put Millions of AI Agents at Risk</title>
		<link>https://firethering.com/badhost-starlette-critical-vulnerability-ai-agents/</link>
					<comments>https://firethering.com/badhost-starlette-critical-vulnerability-ai-agents/#respond</comments>
		
		<dc:creator><![CDATA[Mohit Geryani]]></dc:creator>
		<pubDate>Wed, 27 May 2026 15:26:30 +0000</pubDate>
				<category><![CDATA[Tech]]></category>
		<category><![CDATA[Trends]]></category>
		<category><![CDATA[AI]]></category>
		<category><![CDATA[AI Agents]]></category>
		<category><![CDATA[Security]]></category>
		<guid isPermaLink="false">https://firethering.com/?p=7119</guid>

					<description><![CDATA[One character. That's what it took to bypass authentication on millions of servers running AI agents, MCP tools, and the infrastructure connecting them to user data, email accounts, databases, and in some cases industrial equipment.

The vulnerability, now tracked as CVE-2026-48710 and nicknamed BadHost, was found in Starlette, an open-source framework downloaded around 325 million times every week. If you’re building AI infrastructure in Python, there’s a good chance something in your stack depends on it. 

Starlette is the foundation FastAPI is built on, and FastAPI is what a significant portion of the Python AI tooling ecosystem runs on. 

Researchers say the official severity score doesn’t fully capture how dangerous the bug actually is. A patch was released Friday in Starlette 1.0.1, but vulnerable versions are still running in production systems right now.]]></description>
		
					<wfw:commentRss>https://firethering.com/badhost-starlette-critical-vulnerability-ai-agents/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
	</channel>
</rss>
