Everyone's asking what the hackers stole from the @FBI crime database.
I want to know if they could write to it.
This is the database a cop checks during a traffic stop. Add a fake warrant and you've swatted someone without a phone call. Delete a real one and someone wanted walks free.
When a hacker can read your file it's a breach. When they can edit it, that's a lot worse.
nitter.cf/KobeissiLetter/status/…
BREAKING: The FBI's national crime database has been hacked in a "major cybersecurity breach," per CNN.
The hacked database includes on nationwide information regarding criminal warrants, stolen property, and background checks.
This comes amid several other outages and hacks over the last 10 days.
51
En una semana: aviones en tierra, bancos caídos, fallas en la NSA, el @USDA desconectado… y se va el Director Nacional de Ciberseguridad de la @WhiteHouse.
Nos dicen que son eventos separados. Puede ser.
Pero en ciberseguridad la coincidencia no es una conclusión. Es una hipótesis que hay que descartar.
¿Quién está investigando el patrón, y no solo apagando cada incendio? @ONCD
87
Plot twist in the @Microsoft PyPI hack: the first attempt from AWS got blocked. The second came from a Microsoft-owned network and sailed through.
So the final security check was "are you calling from inside the house?"
And per @wiz_io, the attackers' own AI hacking tool snitched on them, phoning home progress reports to the same server as the stolen data.
Red-team question: does your release pipeline trust the key, or the network it's calling from?
nitter.cf/IntCyberDigest/status/…
‼️ UPDATE: The hackers who planted malware in Microsoft's AI agent toolkit on PyPI pushed it through Microsoft's own release system, with a stolen Microsoft certificate and private key.
Cloud security firm Wiz says their first try, from AWS, was blocked. The second, from a Microsoft-owned network, went through.
How the key was stolen is not yet known. Wiz counted three confirmed thefts of victim data before both versions were pulled within the hour.
The attackers' own AI hacking tool gave part of the operation away, reporting its progress to the same web address the stolen data went to.
1
48
Se cae el banco, se cae Microsoft, se cae Apple… y ¿quién llega primero? Los estafadores. 😅
El miércoles en Univision 34 Los Ángeles: "los bandidos escuchan las mismas noticias que nosotros." Si te meten prisa, es estafa.
arielcoro.com/blog/univision…
73
35 Days of Hacks Until Christmas 🎄
On the 35th day, a hacker gang gave to me: data on 16,000 @Visa, @Mastercard and @jpmorgan customers.
They even ran a 48-hour ransom countdown first. Copycats.
The banks weren't breached. Their vendors were.
Red-team question: how many vendors are holding YOUR customer list?
#DaysOfHacks #Cybersecurity #ThirdPartyRisk
29
14 minutes.
That's how long @HackingDave says an AI agent swarm needed to get past 2FA, steal the data and encrypt an entire company.
Now open your incident response plan and count the steps that wait for a human to pick up the phone.
Your SOC works in shifts. The swarm doesn't.
Defense at human speed is a museum exhibit. nitter.cf/OutFrontCNN/status/212…
Former NSA hacker David Kennedy reveals he’s actively responding to a live breach where an AI “agent swarm” bypassed two-factor authentication, stole data, and encrypted a company's entire system: “This all took 14 minutes. 14 minutes. Never seen anything that fast before.”
Readers added context they thought people might want to know
David Kennedy clarified attackers used valid credentials and a customer app config flaw to circumvent MFA, not that AI bypassed secure 2FA. Segment focused on 14-minute attack speed.
x.com/HackingDave/st…
x.com/HackingDave/st…
ibtimes.co.uk/ai-attack-bypa…
1
32
36 Days of Hacks Until Christmas 🎄
On the 36th day, @RakutenSec gave to me: up to 15.7 million accounts exposed... from a TEST environment.
Test systems are supposed to use dummy data. This one had real customers.
Red-team question: is your test server on Santa's naughty list too?
#DaysOfHacks #Cybersecurity #DataBreach
32
Esta semana CNN y la BBC le sirvieron código malicioso a sus propios lectores.
No los hackearon a ellos. Hackearon la herramienta que usan para probar qué titular da más clics: @Optimizely.
178 organizaciones. Un skimmer de tarjetas y un ladrón de direcciones cripto, servidos desde el CDN del proveedor.
Tu sitio web no es tuyo. Es la suma de todos los scripts de terceros que carga sin que nadie los revise.
¿Alguien en tu empresa sabe cuántos son? @CNN @BBCNews
42
The robotaxi race is being scored wrong.
Everyone's counting cars: @Tesla just registered 200 more Cybercabs in Texas.
Meanwhile California pulled @Waymo off freeways after its cars kept driving into active construction zones. Waymo paused freeway rides nationwide.
The winner won't be whoever has the most cars. It'll be whoever handles the weirdest 1% of the road: cones, flaggers, a cop waving you through a red light.
Fleet size is a vanity metric. Edge cases are the scoreboard.
Who publishes their construction-zone numbers first?
1
29
Your AI agent reads your email.
So does every scammer who can send you one.
@PromptArmor hid attacks in emails, docs and files across 21 AI models. The attacks worked in 34.9% of test cases.
We spent 20 years training employees not to click the link. Now we're handing the inbox to an intern who falls for it about 1 time in 3.
Would you give that employee admin access to your CRM?
nitter.cf/PromptArmor/status/212…
Today we’re releasing FrontierBreach, a new standard for attack success rate on agentic workflows in the enterprise.
Existing benchmarks systematically underrepresent the security risks facing modern AI agents.
FrontierBreach reflects attack scenarios on realistic enterprise workflows, while using sophisticated attack chains that result in cyber consequences (ransomware, data exfiltration, phishing, etc).
1
34
Cybersecurity right now, in 5 lines:
- Attackers exploit bugs ~7 days before a patch exists (Mandiant)
- AI finds holes faster than humans fix them
- Nobody fully understands the systems they defend
- The fixers are booked for months
- No serious pro says it ends soon
Stop waiting for the patch. Plan for the breach.
63
Banks glitching. NSA network down. Welcome to the Hackpocalypse: hackers now have AI doing the hacking for them. 🤖💀 #hackpocalypse
🤖 Made with AI
1
41
37 Days of Hacks Until Christmas 🎄
On the 37th day of hacking, my true love gave to me: a supercomputer breached through its status page.
Japan's Fugaku, once the fastest on Earth, is offline. Per RIKEN, attackers got in via its public "Operating Status" dashboard and left with SSH keys and credentials.
Billions in compute. Owned by the "is it up?" page.
Audit your harmless public pages before Santa's elves do.
57
Un investigador usó Codex de @OpenAI y en cerca de una hora encontró un hueco crítico en Java. Lo bautizaron "Java4Shell". @CISACyber pide parchar YA.
Todos celebran que la IA encuentra fallas en una hora.
Nadie pregunta cuánto tarda tu empresa en parcharlas.
Si el defensor encuentra el bug en una hora, el atacante también. La diferencia ahora no es quién lo encuentra. Es quién actualiza primero.
¿Tu empresa parcha en horas o en trimestres? 👇
81
Microsoft spent Tuesday at Ignite selling the agentic future.
Same day, someone hijacked @Microsoft's official AI agent package on PyPI. Versions 1.26.1 and 1.26.2 stole Azure, GitHub and CI secrets every time Python started. Your code didn't even have to import it.
Your AI agent's first dependency was the attack.
Red-team question: how many "autonomous agents" running in production right now are on a version nobody checked?
100,000+ license-plate cameras. One national database. No court order.
A Milwaukee cop searched his girlfriend's plate 124 times on @flocksafety and typed "investigation" every time. No internal audit caught it.
Red-team lesson: a log nobody reads isn't a control.
arielcoro.com/blog/flock-cam…
1
56
Más de 100,000 cámaras leyendo placas en EE.UU., conectadas a una base de datos e instaladas sin orden judicial.
Un policía de Milwaukee buscó la placa de su pareja 124 veces en @flocksafety. Motivo: "investigación". Ninguna auditoría interna lo detectó.
Lección de red team: un registro que nadie revisa no es un control. Es un diario.
Y ICE no tiene contrato con Flock, pero ha entrado por la puerta lateral: los departamentos de policía.
Lo hablé en @nmasunivision con @VILMATARAZONA 👇
arielcoro.com/blog/flock-cam…
78
Ayer: Univision. Hoy: En Foco con @Lourdes_DelRio. Mismo apagón, otra pregunta: ¿por qué si falla @Chase lo sienten los clientes de otros bancos?
Porque los bancos se esperan unos a otros. Uno se traba y se arma una cadena de dominó.
Tu dinero no desapareció. Pero el SMS como segundo factor, ese sí retíralo. Usa un autenticador. 🔐
arielcoro.com/blog/entravisi…
Did I just prevent a run on the banks? Probably not. 😅
But on Monday on @nmasunivision, in the middle of the @Chase outage, I told people the one thing they needed to hear: your money didn't vanish. Your app just took a nap.
arielcoro.com/blog/univision…
49
¿Acabo de evitar una corrida bancaria? Probablemente no. 😅
Pero el lunes en @nmasunivision, en plena falla de @Chase, dije lo único que la gente necesitaba oír: tu dinero no desapareció. Tu app solo se tomó una siesta.
Los bancos no son islas. Si uno se cae, los demás lo sienten.
arielcoro.com/blog/univision…
94








