Category Archives: Security

Libya Internet Link Cut

At the RSA conference I ran into a Renesys speaker and he introduced himself to me as “we’re the company that broke the Egypt story on the Internet cut”. I asked whether Bahrain would go the same way. He skipped right past the prediction and instead said “It’s not the same right now, not at all. All their routes remain online and stable despite traffic fluctuation.” Of course I can respect this — it is much safer to report current events than predict new ones. Libya, for example

Renesys confirms that the 13 globally routed Libyan network prefixes were withdrawn at 23:18 GMT (Friday night, just after midnight Saturday local time), and Libya is off the Internet. […] We wondered whether anyone would repeat Egypt’s strategy. Tonight, it appears that we have our answer.

Yes, we do. Bahrain still is able to directly upload video.

Stuxnet Failed to Stop or Delay LEU

Three days ago an updated report by the Institute for Science and International Security (ISIS) was published with the following conclusion:

While it has delayed the Iranian centrifuge program at the Natanz plant in 2010 and contributed to slowing its expansion, it did not stop it or even delay the continued buildup of LEU [low enriched uranium]. […] At the time of the attack, the Natanz FEP contained a total of almost 9,000 IR-1 centrifuges. The destruction of 1,000 out of 9,000 centrifuges may not appear significant, particularly since Iran took steps to maintain and increase its LEU production rates during this same period. […] One observation is that it may be harder to destroy centrifuges by use of cyber attacks than often believed.

They suggest that the malware was injected into systems in the supply-chain for Natanz.

Because of sanctions and trade controls, Iran operates international smuggling rings to obtain industrial control equipment, including the Siemens 315 and 417 PLCs. Although foreign intelligence agencies could infect or sabotage these PLCs abroad, they would have far greater chance of ultimately infecting Natanz by inserting Stuxnet in the core of Iran’s supply chain for the centrifuge program’s control systems.

This points strongly to an outsider cut-off from direct site access yet influential, which echoes a CIA method claimed to have caused the trans-Siberian pipeline disaster in 1982. On the other hand, it is said the attackers monitored and continued to modify Stuxnet, almost as if they had inside access and knowledge of their progress:

Symantec has established that Stuxnet first infected four Iranian organizations in June and July 2009. After the 2009/2010 attack, and before Stuxnet’s public discovery, the malware’s operators tried to attack again. Symantec found that in March, April, and May 2010, two of the original organizations were again infected. In May, a new Iranian organization was also infected. Were the Stuxnet operators dissatisfied with destroying only 1,000 centrifuges, or were they encouraged by their success? In any case, they were improving the code’s ability to spread by the spring of 2010, according to Symantec. These improvements undoubtedly sought to enable the program to again breech Iran’s security on its gas centrifuge program and destroy more centrifuges.

The report points out that the level of knowledge required for the attack had to come from a plant insider, but that the attack vector is more likely to have been from an outsider. The blended approach of Stuxnet emphasizes a loss of secrecy in their program, which may significantly affect Iran’s management of their nuclear effort far more than damage to controllers and centrifuges. The objective may have not been destruction but rather to demonstrate the sophisticated level of information leakage.

Cloud SMaaS (Sloppy Meat as a Service)

I just noticed that a reporter had contacted the company in Beijing cited by McAfee as a source of attacks in their Night Dragon report. It was sometimes called Sloppy Night Dragon because the attacks were not well concealed. I started to call it Operation Sloppy Joe because McAfee said the perpetrators were “company men”.

The Associated Press really has the scoop on why it could have been anyone on hosted servers rather than company men:

Song said hackers using his company’s services had an estimated 10,000 “meat computers” controlled remotely without the owners’ knowledge. He said “yes” when asked whether such activities might be improper but he said Chinese authorities never have contacted him about them. He hung up the phone when a reporter asked for other details.

He’s a service provider of meat computers used in Operation Sloppy Joe?

If only I had known sooner…the world is being attacked by “sloppy meat as a service”.

But seriously, that could be a really bad translation.

肉 (pronounced “zh-rou“)

noun
1. meat
2. flesh
3. beef

Maybe he said something like I provide “beefy” computers, or “meat” is Chinese slang for big and powerful if less literally translated. I doubt he said he has fleshy computers, although even that has a plausible translation — easily commandeered.

I bet the provider in Beijing even offers sub-levels of Chinese SMaaS:
1) Rare – infrastructure (RSMaaS)
2) Medium – platform (MSMaaS)
3) Well-done – software (WSMaaS)

Rumor has it that #3 is available with BROCCOLI (Binary Rootkits and Computational Compromise of Logical Information)

Crowd Attribution and Riots

Tuesday I gave a presentation where I described the phases of Operation Ajax, the CIA plot to overthrow the democratically elected leader of Iran in 1953.

One of the key phases involved fake mobs. The US gave money to people, apparently both for and against Prime Minister Mossadegh, to create the appearance of chaos and the need for order from military power.

After the presentation I started to notice news of growing protest crowds in Wisconsin. Today pro-rally new sites say that angry crowds have grown into the tens of thousands.

Madison Police Chief Noble Wray, a veteran of 27 years on the city’s force, said he had has never see a protest of this size at the Capitol – and he noted that, while crowd estimates usually just measure those outside, this time the inside of the sprawling state Capitol was “packed.”

On Wednesday night, an estimated 20,000 teachers and their supporters rallied outside the Capitol and then marched into the building, filling the rotunda, stairways and hallways. Chants of “What’s disgusting? Union busting!” shook the building as legislators met in committee rooms late into the night.

Although it might be hard to get to attribution with public crowd numbers, as mentioned above, all Madison schools were shutdown this week as teachers collectively called in sick. You can’t argue with those numbers. Empty school rooms are easily counted.

The Governor of Wisconsin has responded by calling protesters a minority and threatening them with action by state militia.

In an interview with Fox News’ Greta Van Susteren on Tuesday, Walker played down the number of protesters. Van Susteren observed that tens of thousands of residents have turned out to protest, saying, “I don’t think Madison has seen a protest like that in quite some time.” Walker replied, “In the end though, you’re still talking about 5.5 million in the state. You’re still talking about a couple of hundred thousand state and local government employees. So sure, you’re going to have a few riled up about this, there’s no doubt about it.”

Tens of thousands in protest and schools shutdown seems significant to me. Walker seems to emphasize that he has a formula to estimate how many people (or which key people) must oppose him before he will admit a loss of popular confidence in his position.

While this was unfolding, I caught an interesting article on the HBGary breach.

Just today I was listening to Stand Up with Pete Dominic on XM’s POTUS channel. He was talking about the Wisconsin labor attack and how he had seen a lot of people email and contact the show in support of the Teachers there. Then he added a “but”: “I’ve also seen a lot of anti-labor people on Twitter…”

Really? I thought. How do we know if those are real people? Twitter has to be the easiest thing to fake and to automate with retweets and 180 characrer max sentences. To the extent that the propaganda technique known as “Bandwagon” is an effective form of persuasion, which it definitely is, the ability for a few people to infiltrate a blog or social media site and appear to be many people, all taking one position in a debate, all agreeing, for example, that so and so is not credible, or a crook, is an incredibly powerful weapon.

That is related to the HBGary breach because it turns out that this is a project mentioned in a leaked internal email message:

According to an embedded MS Word document found in one of the HB Gary emails, it involves creating an army of sockpuppets, with sophisticated “persona management” software that allows a small team of only a few people to appear to be many, while keeping the personas from accidentally cross-contaminating each other. Then, to top it off, the team can actually automate some functions so one persona can appear to be an entire Brooks Brothers riot online.

The 1953 example again comes to mind. That is exactly like the model actually used by the CIA to generate chaos and conflict in the streets after the Iranian PM was arrested and soldiers were deployed.

What we are seeing is the steady erosion of attribution, which perhaps is best seen in the measure of crowd events and growth statistics. When a site says they now have 500 million users; a user to real person ratio could be as high as 100:1 or more.

The loss of attribution to an HBGary system is not a development that comes out of the blue. It has been a result of users defending themselves against weak privacy controls.

Tuesday night I met with a group of security professionals and tossed this idea out as a security trend — new generations will not only further erode attribution, they will be exceptionally adept at it as a form of self-preservation. They will be better at privacy than older generations because they will enter into a world actively debating the best way to achieve anonymity and repudiation. They will learn from and adapt around privacy mistakes made by others before them. As older generations continue to wonder why they can not trust someone online like they do offline, and how they might fix the issue, new generations will be far more prepared to see a lack of attribution and weak trust as the norm.