惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

TaoSecurity Blog
TaoSecurity Blog
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
F
Fortinet All Blogs
Cisco Talos Blog
Cisco Talos Blog
D
Darknet – Hacking Tools, Hacker News & Cyber Security
S
Secure Thoughts
美团技术团队
雷峰网
雷峰网
Hugging Face - Blog
Hugging Face - Blog
博客园_首页
C
CXSECURITY Database RSS Feed - CXSecurity.com
Engineering at Meta
Engineering at Meta
人人都是产品经理
人人都是产品经理
月光博客
月光博客
T
Tor Project blog
P
Privacy & Cybersecurity Law Blog
Recorded Future
Recorded Future
I
Intezer
博客园 - 【当耐特】
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
GbyAI
GbyAI
罗磊的独立博客
V
V2EX
Google DeepMind News
Google DeepMind News
D
DataBreaches.Net
Last Week in AI
Last Week in AI
T
Tailwind CSS Blog
www.infosecurity-magazine.com
www.infosecurity-magazine.com
A
About on SuperTechFans
Scott Helme
Scott Helme
Vercel News
Vercel News
Spread Privacy
Spread Privacy
T
Threat Research - Cisco Blogs
Recent Announcements
Recent Announcements
Hacker News: Ask HN
Hacker News: Ask HN
C
CERT Recently Published Vulnerability Notes
G
Google Developers Blog
B
Blog
博客园 - 叶小钗
WordPress大学
WordPress大学
博客园 - 聂微东
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
Jina AI
Jina AI
IT之家
IT之家
C
Cybersecurity and Infrastructure Security Agency CISA
P
Palo Alto Networks Blog
小众软件
小众软件
博客园 - Franky
Microsoft Azure Blog
Microsoft Azure Blog
AWS News Blog
AWS News Blog

The Register - Special Features

Troops’ phones gave away location data to foreign adversaries Qualcomm picks bad time to pitch a $300 laptop platform AI agents get their own phone directory built atop DNS Carnival confirms ShinyHunters cruised off with 6M customer records after April breach Google engineer accused of turning Year in Search secrets into Polymarket payday Are we human? India's cyber agency sets clock at 12 hours to tackle exploited bugs as AI turns up the heat Broadcom gets early start on WiFi 8 with next-gen wireless routing kit Are we human? Microsoft Excel champ proves he still has the formula Anthropic co-founder hallucinates ghost in the machine Anthropic co-founder hallucinates ghost in the machine NASA plans Moon Base buildout with rovers, drones, cargo landers MyPillow must decide whether to be firm or soft as ransomware crims demand pay Starship shows it can deploy satellites, but Moon mission clock still ticks Huawei's chip law looks less like Moore and more like marketing Experts pour cold borscht on Farage's Russian hack claim Logitech unveils a cushioned mouse for all-day use AI eyes scanning for bugs create a worrisome Linux security trend A Russian speaker and jailbroken Gemini went on a hacking spree and emptied at least one MAGA victim's crypto wallets AI datacenter boom collides with US grid reality Media giant settles for $930k amid user-snooping allegations AT&T sues to ditch Cali copper phone lines to save billions FBI warns of Kali365 as device code phishing soars Techie claims Trump Mobile website was leaking thousands of people's data BOFH: Vibe-coded solutions arrive for problems nobody has Dems slam Trump for making cybersecurity hold out the tin cup while splurging on ballroom and Jan. 6 'slush fund' Google explains how it will infuse ads into AI answers AI is getting pricey, but relief is coming, but not for you Deus ex machina: Half of US Christians trust AI's spiritual advice Attackers spill plaintext passwords of 46k Myspace93 users after 2021 breach Apple adds AI smarts to Voice Control, VoiceOver and Magnifier ahead of Accessibility Day Microsoft open-sources agentic AI safety tools OpenAI wants upfront cash for guaranteed AI capacity Fedora: Microsoft is all aboard, but Deepin is dumped Bye-bye, Gemini CLI; Google nudges devs toward Antigravity Plex appeal fades as Lifetime Pass jumps to $750 AI sackings reach New Zealand, which will use it to eject 14 percent of government staff Anthropic’s Stainless steal tightens grip on AI dev tooling Are we human? Google touts tokenmaxxing, huge capex, and AI agents at I/O America's top cyber-defense agency left a GitHub repo open with with passwords, keys, tokens – and incredibly obvious filenames America's top cyber-defense agency left a GitHub repo open with passwords, keys, tokens – and incredibly obvious filenames Shadow AI invades the workplace, up 4x in the last year Microsoft refreshes Surface for Business lineup, starts AI PC upsell at $1,499 Broadcom finds a VMware customer willing to stick around: London Stock Exchange 468k records allegedly stolen from Portugal’s postal carrier Baidu says the quiet part out loud – you can’t build AI infrastructure, so clouds can cash in Shai-Hulud copycat worm infects yet another npm package Uncle Sam's next big super might not use GPUs Are we human? Datacenters slurping up so much juice they boosted prices 75% in largest US energy market MPs want social media treated more like unsafe toys than harmless apps Cerebras’ wafer-scale AI bet delivers blockbuster IPO Nobody believes the 'criminals and scumbags' who hacked Canvas really deleted stolen student data Anthropic tosses agents into the API billing pool Jen Easterly, cybersecurity's 'relentless optimist,' hopes feds come back to RSAC next year Jen Easterly, cybersecurity's 'relentless optimist' Smooth criminals talking their way into cloud environments, Google says Voice phishing skyrockets as smooth crims talk their way in RSAC 2026: Uncle Sam backs out, AI agents everywhere RSAC 2026: Uncle Sam backs out, AI agents everywhere Decoding Nvidia's Groq-powered LPX and the rest of its new rack systems A closer look at Nvidia's Groq-powered LPX rack systems Nvidia slaps $20B Groq tech into massive new LPX racks to speed AI response time Nvidia slaps Groq into new LPX racks for faster AI response AI Burning Man happens next week – what to expect at Nvidia GTC 2026 Nvidia GTC 2026: What to expect at AI Burning Man Unaccounted-for AI agents are being handed wide access Unaccounted-for AI agents are being handed wide access Google to foist Gemini pane on Chrome users Google to foist Gemini pane on Chrome users Yes, you can build an AI agent – here's how, using LangFlow How to build an AI agent using LangFlow Clawdbot becomes Moltbot, but can’t shed security concerns Clawdbot becomes Moltbot, but can’t shed security concerns Gartner questions if Salesforce AI will stay all-you-can-eat Gartner questions if Salesforce AI will stay all-you-can-eat Claude supports MCP Apps, presents UI within chat window Claude supports MCP Apps, presents UI within chat window Cursor is better at marketing than coding Cursor is better at marketing than coding Feds skipping infosec industry's biggest conference, RSAC AI is rewriting how power flows through the datacenter All aglow about DCs, investors launch $300M at microreactor startup Radiant bags $300M-plus to commercialize its microreactors Why do bit barns keep bumping up our bills, Senators ask DC operators Senate trio questions DC operators over rising energy costs Building the AI factory datacenter Delays? What delays? Oracle insists its $300B cloud contract with OpenAI is on track Oracle insists its $300B contract with OpenAI is on schedule Salesforce willing to lose money on AI to lock in customers Salesforce willing to lose money on AI to lock in customers Galactic Brain space datacenter coming in 2027, pledges startup Aetherflux Galactic Brain space datacenter promised in 2027 Activist groups urge Congress to pause datacenter buildouts Activist groups urge Congress to pause datacenter buildouts Bezos-backed Unconventional AI addresses datacenter power Bezos-backed Unconventional AI addresses datacenter power AWS re:Invent keynote: Matt Garman bores, then thrills
They read the scroll thing! AI helps decipher ancient document charred by Vesuvius
Brandon Vigliarolo · 2026-06-26 · via The Register - Special Features

offbeat

'Having certainly strained ourselves to the utmost through research and learning, we will no longer be inferior to them,' reads a scroll virtually unwrapped with the help of AI

A sealed scroll from the Roman town of Herculaneum, which was destroyed by Mount Vesuvius' eruption nearly 2,000 years ago, has finally given up its secrets, thanks to a combination of machine learning and high-resolution CT scans.

In 2023, researchers managed to decipher a few words from among the char and ash that make up the bulk of the scrolls. Some of those same prize-winning researchers recovered more passages from one of the scrolls, PHerc.Paris.4, netting them the $700,000 grand prize from the Vesuvius Challenge contest in early 2024. 

Fast forward two more years, and those grand prize winners are now part of the Vesuvius Challenge team that managed to read the surviving portion of a rolled scroll end-to-end, as the VC team shared in a Thursday announcement and detailed in an accompanying paper [PDF]. 

According to the research paper, the ability to make out the entirety of the scroll was thanks to high-resolution phase-contrast X-ray microtomography performed at the European Synchrotron Radiation Facility in France - an improved imaging technique over prior methods used to capture prior images that were analyzed in the prize competition. 

That wasn’t all, though: They say that much of their work succeeded because of a new "workflow" they developed to scan scrolls, detect ink on charred papyrus, virtually "unroll" the scrolls by modeling their deformed surfaces, and preserve those surfaces digitally, allowing machine learning models to identify letters across an entire scroll rather than just isolated patches.

"The key transition marked by the present work is therefore from exceptional local recovery to systematic scroll-scale recovery," the team wrote. In other words, provided they can account for the particularities of the hundreds of sealed scrolls recovered from Herculaneum's Villa of the Papyri, the world's only surviving intact library from antiquity, this could mark the beginning of an explosion in new material for historians.

So, what did it say?

PHerc.Paris.4 wasn’t at the center of this breakthrough either, though they did have some exciting news to share on that front that we’ll get to. Instead, the breakthrough centered on PHerc. 1667, a previously unread rolled scroll whose preserved text was read continuously from end to end for the first time. The work appears to be a treatise on Stoic philosophy focused on ethics - a favorite subject of Zeno, Marcus Aurelius, Epictetus, and their intellectual fellows.

“Having certainly strained ourselves to the utmost through research and learning, we will no longer be inferior to them in any respect,” a passage from the latter part of the scroll reads, “accomplishing in like manner the things that befit them and possessing the same practical wisdom as they.”

Quite a fitting bit of ancient wisdom to be the first to see the light of the modern world. 

While the team digitally unrolled the scroll, detected its ink, and transcribed the preserved text from end to end, portions of the original PHerc. 1667 were lost long ago during earlier attempts to physically open the scroll, before archaeologists had access to sophisticated X-ray imaging and AI-assisted analysis.

“Earlier attempts to open it by hand — in the nineteenth century, and again in 1969 and the 1980s — destroyed its outer layers,” the Vesuvius Challenge team said, noting that only an 8 cm-high core remains of the original scroll, which originally measured between 19 and 24 cm in height when standing upright. Nonetheless, “it is the first time the preserved text of a rolled Herculaneum scroll has been read continuously, end to end, rather than in isolated words or patches,” the team said. 

In addition to the full reveal of what’s left of PHerc.1667, the team also managed to pick out some information from a couple of other scrolls using their new workflow. One, PHerc.139, was determined to be a copy of book eight of epicurean philosopher Philodemus’ treatise On Gods, meaning scholars can expect to know what they’re looking at once the scroll is fully digitally unrolled. 

A comparison between the original scans of PHerc.Paris.4 used in the prize competition (top), and a scan taken using the new method

A comparison between the original scans of PHerc.Paris.4 used in the prize competition (top), and a scan taken using the new method
Source: Vesuvius Challenge/Giorgio Angelotti, et al.

The second concerns, as mentioned above, PHerc.Paris.4. The new higher-resolution images taken for this latest experiment make the words on the scroll directly visible for the first time, meaning that there’s no need to rely on algorithmic detection of individual words and phrases from CT scans. Most crucially, the new scans of Paris.4 perfectly matched what the grand prize team made out several years ago, providing independent confirmation that the prize went to the right team. 

There are still challenges to meet in unwrapping and deciphering the rest of the ancient library, with the team calling out geometric challenges in surface prediction that can render an unrolled scan unreadable, and radiometric challenges that make ink identification difficult, as ancient recipes were inconsistent. Still, it’s a massive leap forward and the team believes the X-ray and machine learning workflow they’ve developed is ready to scale. 

“The thoughts of the ancient world, sealed in darkness for two millennia, are coming back into the light — a whole scroll at a time,” the Vesuvius Challenge team said. I, for one, can’t wait to see what ancient secrets they discover next. ®