Intel Is Maintaining Legacy Technology for Security Research


Intel’s issue reflects a wider concern: Legacy technology can introduce cybersecurity weaknesses. Tech makers constantly improve their products to take advantage of speed and power increases, but customers don’t always upgrade at the same pace. This creates a long tail of old products that remain in widespread use, vulnerable to attacks.

Intel’s answer to this conundrum was to create a warehouse and laboratory in Costa Rica, where the company already had a research-and-development lab, to store the breadth of its technology and make the devices available for remote testing. After planning began in mid-2018, the Long-Term Retention Lab was up and running in the second half of 2019.

The warehouse stores around 3,000 pieces of hardware and software, going back about a decade. Intel plans to expand next year, nearly doubling the space to 27,000 square feet from 14,000, allowing the facility to house 6,000 pieces of computer equipment.

Intel engineers can request a specific machine in a configuration of their choice. It is then assembled by a technician and accessible through cloud services. The lab runs 24 hours a day, seven days a week, typically with about 25 engineers working any given shift.

Slashdot thread.

Posted on November 30, 2021 at 1:28 AM34 Comments


Doug November 30, 2021 2:47 AM

only a decade old? In our school office there are many computers from the core2duo era, 15 years old.

Peter A. November 30, 2021 3:00 AM

Cool technology museum… I would love to play around 🙂 Ok, ok, I know, it’s a closed research lab, sigh.

There are some private collectors over here who make their limited collections of vintage computers and stuff (e.g. pinball machines!) available to the public for a moderate fee – but public institutions in my wider area apparently do not show interest in “modern” equipment. You could see old steam machines, locomotives (some still running), cars, trams, old printing hardware (up to a fairly modern offset machine, but no DTP), etc. etc.

In the not so distant past (my university days) I dreamed about creating a computer museum, but did not have money or influence to do so (and still do not have – it’s too late anyway). I personally saw all these VAXen etc. go to scrapyard… stupid bureaucracy…

EEC November 30, 2021 6:02 AM

Sound almost like the reverse of what the story says. They are gathering secure pre-“Intel ME” CPUs so they themselves aren’t the future target of their own spyware (Spyhardware? Spyfirmware? Are those words? I guess now they are!) 😉

Otherwise “about a decade” seems way to little. My newest computer is 9 years old, and my oldest in common use is about 15 (with “common use” I mean running modern linux operating systems and used for normal everyday stuff, I’m not counting my vintage computers with ages of 25-45 years old).

The decade 2010 to 2020 is nothing like, say, 1980-1990 or 1990 to 2000 and there is not any substantial difference between an i7 with 4-8 cores from 2010 and something now. Not like going from Intel 8086 to 80486DX or 80486DX to 1 GHz Athlon in the past two decades.

As Doug said, never mind core i7, it must be tons and tons and tons of core 2-CPUs “out there” in offices. Even P4s, Ahtlon XPs and the like. I know embedded systems running 386 and 486 to this this day (And I don’t mean newer SoC 486 manufactured today, but old socketed stuff.)

John November 30, 2021 6:37 AM


It is easier to get SIP audio to run smoothly on a old Microchip 8 bit cpu than on a ‘modern’ Intel cpu?

Try it yourself!

Now let’s hear the justifications for that :).

My favorite is “all software has bugs.”…. really??


EEC November 30, 2021 6:42 AM

@John: Who/what exactly are you answering? I can’t see Microchip PICs, audio or bugs mentioned anywhere in this thread, in the slashdot thread or the linked article.

Clive Robinson November 30, 2021 7:04 AM

@ ALL,

At only a decade old this is “modern” and will be insecure by default due to Intels ME and Ring -3 nonsense.

As I’ve noted in the past @Nick P one or two others and myself had long debated when the Security “Best Before Date” on PC hardware was.

@Nick P favoured 2005 (now known to have been insecure) and I favoured “last century” for Intel products (still “assumed” to be secure for early Pentiums, 486s and earlier).

It’s why I have getting on for a dozen 8088, 286/386 and 486SX/DX machines running currently. That inportantly have “No Flash ROM” on the motherboards or interface cards, no RF Data-OTA products or even modem cards. All but a couple do not have any built in I/O other than mouse/kby and serial/parallel ports. Some only have floppies both 5.25 and 3.5 and old style “card readers” with a couple of QIC and DAT tape drives and oh NE2000 “cheapernet” network cards and some faster 3Com Etherlink cards that have PXE BOOT PROMS (saves on mechanical wear on old hard drives and floppies).

I have the parts to make probably another five or six PC’s but cases and especially power supplies are the problem.

Sometimes finding info is hard after all if you’ve only heard of UFEI the likes of PXE, BOOTP, TFTP, etc is something you have not heard of so do not know. So you can not search for info on them.

This is a fun list of things to help you find terms you did not know, but will do so you can then search fruther on,*nix

But for those that want “Hardware security” but can not get (or aford) last century PC tech…

One solution is modern “microcontroler” development boards. You can get single chip microcontroller chips for less than $2 that will run early versions of Unix from last century. As a hobby or even just a challenge quite a number of people have ported early BSD onto such devices.

For example RetroBSD onto Microchip PIC32 microconyrollers,

No you will not get the modern experience, it will all be “Command Line Interface”(CLI) but appart from graphics, you can do a lot of the day to day stuff.

But there is also PC/104 embedded “Idustrial Control Systems”(ICS),

That are based on the PC-XT 8bit and PC-AT 16 bit backplanes which are simple to “design to”,

Whilst attention is now shifting to PCi/104 PCI based the old 8/16bit designs are still quite easily found (due to the fact ICS installations generally have long life times with millions of dollars invested in them).

But also aside from very useful multiple serial ports and the likes of CAN-Bus and other hardware I/O. There are interesting “industrial cases” readily available, some of which are designed to meet very tight EMC specifications for test equipment etc. Thus you can buy off the shelf “Faraday Sheilded” fanless systems for a supprisingly low cost.

Oh and they are “easy to build into safes”… In the US for instance LEO’s have pushed computers out of the “out of sight” protection that documents etc have, by being in locked draws etc. A computer especially a non-standard looking computer built into a locked safe is going to be hard to argue is not “out of sight”. Further those “Fire-Resistant” safes give you a number of options NAS boxes etc do not as easily give.

Ted November 30, 2021 8:12 AM

Good for Intel.

I hope they get the money they need to regain a place in chip manufacturing and technology market.

With supply chain bottlenecks, it looks like Congress is keener to lavish money on companies that want to keep US-based businesses running.

Pat Gelsinger, the CEO of Intel, says, “we should be striving to bring the US to 30% of worldwide semiconductor manufacturing.” The article says the US is at 12% today.

Ted November 30, 2021 8:23 AM

Also, someone on Slashdot commented that it seems contradictory to have a centralized, secure location where tests can be run from anywhere in the world.

I think that’s a good point. However, if the Long-Term Retention Lab has the ability to recreate unique configurations for testing at least it won’t expose a vulnerability on the scale of a class break.

John November 30, 2021 9:58 AM


I agree that ‘flash’ memory is a real problem!! For me, RetroBSD runs well on PIC32MX. PIC32MZ would be a very good solution except that the chip seems to have some serious bugs. Eventually I was able to get it to load my own software reliably, but actually using its serial ports seems to have serious clock issues.

It is hard to design with a chip when even simple serial ports have issues!

My point about SIP is that CPU speed is NOT the problem. The problem is that the low level driver paths are often effectively blocked. Just poor software design. Not a real problem.

It would be nice to find a Linux version that really does SIP well. I expect a lot of other things would also run well.


Freezing_in_Brazil November 30, 2021 10:21 AM

I have a couple of old machines in my office [486 to Pentium4] from the SasC era[1].

They are fully operational and wonderfully air-gapped. Besides research, I really like some of the old programs [e.g. AutoCAD 2000].

[1] Software as something Cool

Clive Robinson November 30, 2021 11:48 AM

@ John,

It would be nice to find a Linux version that really does SIP well.

You probably will not, it’s not the right form of multitasking.

Many many moons ago I developed an audio and video encryptor product that just about hung in on a dual pentium computer (I still have) that cost I guess about the equivallent of four months average income.

I ended up having to write my own device drivers to get it all to work, and it was not expandable with improving hardware. Which is a clue as to what the real underlying problem is.

The thing is that real problem is that commercial OS’s are realy crap at doing certain things. They realy can not multitask except coarsely and then increadibly badly for various reasons.

To get the acurate timing and fine time slices you need you have to use either a Hard RTOS or write the whole darn thing in assembler as the only task on the CPU apart from the clock generator interupt.

I wrote a ~150 page paper on how to do background and foreground processing on 8bit micros used in telephony projects back late last century for a company I worked for that specialised in the design of consummer and commercial communications equipment including both analog and digital two-way radios, cordless, DECT, Cellular and Mobile phones, and AX25 and similar radio modems and packet nodes.

Whilst I don’t own the rights on that work due to the joys of employment so can not share (not sure who oens them these days). I can say that the conclusion was Hard RTOS with very fast time slices (ie north of 10kHz, 16k being just about acceptable) and very few processes. Further that all low level I/O to be done by interupts with tiny tiny handlers and circular buffers. These pulled/pushed to the actual device drivers that were driven by the timer interupt giving them one or more time slices. Likewise the actual processes got time slice driven. As you can appreciate even with 32k/Sec time slices there were hard limits on just what you could do in software, the likes of “bit-banged serial comms” were not the way to go… But by following the rules you could get an 8bit processor to take digitized audio encrypt it and send it on it’s way in the right time slot of an MPX system with a just over 2Mbit/Sec clocked data rate.

Clive Robinson November 30, 2021 1:35 PM

@ Freezing_in_Brazil,

Software as something Cool

I prefere,

“SaS-CWHE” Software as something Compatable with human expectations

Back in the Big Iron and Unix Mini days, when the hight of status was to have one of those amber/orange coloured TTY’s not that harsh green. Something changed… status became more than mear soft colour,

If the IT Gods had blessed you, you had a DEC Rainbow[1]…

With Graphics and harddrive upgrades you could rin CP/M on the Z80 and if you were sensible and also had the V20 upgrade PC-DOS for 86 and later MS-DOS…

By reverse engineering the ROMs we worked out how to make a patch to get it to do both simultaniously on two seperate displays… An upgrade that with hind sight might have been quite valuable… However it ment you could be running a nice game on the 86 and Colour Monitor and a spread sheet / editor / assembler on the Z80 monochrome display a little keyboard short cut would blank the colour display… Now what hacking-nerd is going to give up such an advantage, for just a few pieces of silver, that would go in the “Big Boss” pocket anyway…

The odd thing about the Rainbow was it’s floppy drives… Known less than entusiastically as the “toaster” the disks had to go in upside down with respect to each other (DEC used a common drive motor and if I remember correctly head drive as well, oh and their mind boggling expensive pre-formatted floppy disks).

[1] Less than a decade later it was the NeXT Cube… That would even have that cute secratary outside the Big Boss office think you a God in disguise 😉

SpaceLifeForm November 30, 2021 2:52 PM

@ John

Note SIP is not secure, and getting SIPS to interoperate is likely a non-starter, so I would not entertain this route.

John November 30, 2021 3:12 PM


Yup. That was my point. I have a version of RetroBSD for which BSD was the background process and the ‘idle loop’ and fast interrupt drove the ‘real-time’ stuff. Works great.

It still surprises me that given all the versions of Linux, this simple solution to the underlying problem and circular queues is not in some Linux version!

Anyone who has written assembly math routines dreams of have a ‘Linux’ or ‘bsd’ or forth running in the background for these dreary but not time critical tasks!!!

The only real remaining problem for RetroBSD is the need for kernel loadable modules so the kernel can be left alone while the drivers, etc. are being peeked and poked and otherwise gotten to work.

RetroBSD on a plain PIC32MX otherwise seems pretty solid :).

Now just add some VMS juice :).


ResearcherZero November 30, 2021 3:27 PM

The federal government has some systems like that, and local governments. This year a couple of departments detected simulated attacks on their networks, that were conducted as part of the yearly audits. It only took a few weeks this time around for them to notice.
Definitely an improvement. Someone is actually checking the logs (where possible), or at least some departments are complying with the reporting requirements.

Clive Robinson November 30, 2021 5:10 PM

@ SpaceLifeForm, John, ALL,

Note SIP is not secure, and getting SIPS to interoperate is likely a non-starter, so I would not entertain this route.

No SIP is not secure and was not designed to be, in fact it was designed to be as simple as possible.

The reason was IPv6 SWAN thinking at the time it was designed.

It was assumed that security would be in one of only two places,

1, At the bottom of the presentation level for Application P2Psec.
2, At the network intetface level for Link H2Hsec (VPN).

It’s actually the logical and sensible way to do things and why IPv6 was oh so much more than an Ipv4 replacment with near unlimited addresses.

As for why SIP has interoperability issues, where would you like me to start?

I guess the first thing was that “Every one jumped the gun” with “Favourit featurism” kind of like Microsoft’s famous “Extend, Embrace and, Slaughter” policy.

In telecommunications circles there is a truism that is so constant it’s a reality in it’s own right,

“Standards are like toothbrushes. Every one agrees that you should have one and everyone agrees you should use it religiously. But even when they don’t have one of their own nobody else ever wants to use yours.”

Clive Robinson November 30, 2021 5:42 PM

@ John,

It still surprises me that given all the versions of Linux, this simple solution to the underlying problem and circular queues is not in some Linux version!

Well… The direction Linux is going is into “user land I/O” and “Ever increasing kernel size”. Neither of which favours low power microcontrolers. After all you can not run a 2MByte kernel with all it’s bells, whistles, and plug-ins in 1/8th of a meg of RAM… Let alone have any space for user-land I/O. Which brings up another issue, how do you get effective User I/O in a strictly time sliced environment where you do not have access to the interupts for security reasons…

These problems are far from insurmountable, and some one will say “but you can do…”, the thing is we don’t. Why we don’t I do not know, but I know who you should ask, but he’s got a bit of a reputation for not being able to see beyond certain scope restrictions.

If you want a historic reason, look at his early comments then finally capitulation over the use of the Intel On-Chip “alleged” TRNG.

With regards,

RetroBSD on a plain PIC32MX otherwise seems pretty solid 🙂

Which hardware are you using, I’ve a number of Microchip 16bit development boards from around 2008 and I could not get things working IO wise as things like the “PicTails” did not work as advertised, though they had never given me any problems when developing quit complex code in C for a location based advertising system for London Black Cabs that used 3G/4G to download and store new add videos etc and display them on LCD screens, not just on top of the cab but in the passenger compartment as well (short adds on top, longer adds with sound for the captive audience in the passenger compartment).

The big advantage for RetroBSD as far as I was concerned was not that it was 16bit, but BSD 2 unlike 4 did not require or even support a MMU.

SpaceLifeForm November 30, 2021 7:03 PM

@ Clive

Which brings up another issue, how do you get effective User I/O in a strictly time sliced environment where you do not have access to the interupts for security reasons…

It is simple, you run inside a VM inside a cloud that you do not control. 🙂

John December 1, 2021 2:15 AM


Which PIC32MX hardware are you running on?

One runs on Digilent Max32. But the other 3 are breadboards built from chips. They are ‘real’ breadboards that only a mother could love!! Mount SD cards and boot from them, etc. I start my own serial debugger and boot to BSD from there which allows peek and poke into the running BSD system. System/user issues are the only real annoyance and in my opinion are basically irrelevant for embedded stuff anyways!!

The only real problem I encountered was/is the requirement for a very low ESR cap to get the clock oscillator to start initially. I also built a custom JTAG loader so I could recover from misconfiguration and discovered that an external clock is REQUIRED to recover from some states. That discovery took a LOT of work!

One breadboard is marginal enough that I use a heat gun to get it to boot if it is cold. Once running it runs for long periods with no problem.

I can connect to it VIA RS232 from a Linux host and debug, upload and download code, etc.

Since my background is telephony I am into hardware that ‘keeps running’. That is restarts itself with a keep alive timer, etc. All old stuff as you doubtless know in the telephony world.

We can all thank Bill Gates for the reset button on the IBM PC 🙂

I hope this answers your real questions?

Regards SIP being non-encrypted. Who cares?? The wired telephone world was successful because it was and still is very simple and easy to understand.

Kinda like the early UNIX apps! IRC just works for example. Who needs or want the Google Giggle?

Just my humble opinions….

You can probably fill most of the rest 🙂 :).


Peter A. December 1, 2021 3:46 AM

@Clive Robinson: “the height of status was to have one of those amber/orange coloured TTY’s not that harsh green.”

This reminds my of an “adventure” from the past. There was a computer lab at the university I used to frequent. There were seats with nice “paper white” TTYs connected to a local *nix box, and a few old fiendish-green “refrigerators” with clunky keyboards connected to a VAX eight floors below (RS-232 serial over 20 mA current loop). The *nix seats were almost always occupied, and sometimes queues formed. Waiting with my friend for someone to finally get tired hacking or have go to the loo, we decided to kill some time by lazily hacking at the “green ones”. After an hour or three, a colleague at the *nix TTY called me over to help with his code. I looked at the screen and to my surprise, it did not look paper-white but pinky-white. Hmm, is the CRT failing? I looked at another one – same thing. I looked at the pale sky outside… it was pink too!

‘Tom!’ I shouted to the friend at VAX TTY ‘what colour is the sky?’. He looked through the window and with a surprised look on his face he said: ‘pink!’. The stupid green phosphor recalibrated our brains’ colour perception! The sensation lasted for a good few hours if not more (hard to say as the evening has come soon), luckily the next morning we were both ok.

6449-225 December 1, 2021 5:02 AM

Ok, the quantum substrate now demands an extract from the Jargon File –

Tom Knight and the Lisp Machine

A novice was trying to fix a broken Lisp machine by turning the power off and on.

Knight, seeing what the student was doing, spoke sternly: “You cannot fix a machine by just power-cycling it with no understanding of what is going wrong.”

Knight turned the machine off and on.

The machine worked.

Clive Robinson December 1, 2021 6:36 AM

@ Joun,

That Digilent Max32 board.

I know originally used PIC32MX not parts so had no MMU. Then the PIC32MZ came out which is in many ways “pin for pin compatable”.

I’ve been told –but have not verified– that there are MZ parts on some of those boards now which means they have an MMU so moving from BSD 2.11 of RetroBSD to BSD 4.4 is probably possible, and some apparently have attempted it

(I’ve been told the 4.4 has been forked to it’s own project name).

From my point of view BSD 4.4 has quite a few advantages over 2.11 in terms of not just licencing but maintainability especially on the toolchain over 2.11. But in turn 2.11 has it’s own advantages especially in embedded systems where MMU, Virtual Memory and how it simplifies the tool chain and improves certain security does not realy matter, or is actually a hinderance and takes up scarce resources needlessly.

Freezing_in_Brazil December 1, 2021 6:59 AM

@ Clive

Back in the Big Iron and Unix Mini days, when the hight of status was to have one of those amber/orange coloured TTY’s not that harsh green. Something changed… status became more than mear soft colour,

Those were the days, my friend. I was 16 when I wrote my first Fortran program [the Fibonacci sequence, obviously]. The machine was a VAX. I spent a week with a mesmerized look at that 80-column form with my printed masterpiece. I never had an amber/orange terminal, though.

John December 1, 2021 8:28 AM


Yes. I have 4.4 running on a MZ as well. Not nearly as polished as on MX. Having more real ram would be very nice but there seem to be other more serious problems with the current MZ chips. The serial port for the version that works is done VIA USB! Good luck getting their ‘real’ UARTS to really work. Or in getting licensed USB code!!

And with many other micros RAM cannot have executable code!!

A fixed MZ would be nice. But that would still not big enough to run Linux!

All the action is in Linux for the reasons mentioned. But Linux needs real RAM and HUGE amounts of it!! Not exactly what you want for a real-time system to say the least.

If I could, I would revert to 6502 with 512K of RAM 🙂 :). The 2mhz 6502 product I have just works.

Lots of fun :).


Clive Robinson December 1, 2021 12:24 PM

@ John,

If I could, I would revert to 6502 with 512K of RAM 🙂


There are 6502’s out there that run at nrarly 100MHz and have 2Mbyte of Flash ROM and could have 2Mbyte of ram all on a single piece of silicon.

The thing is that they are “hidden” inside the likes of “System on a Chip”(SoC) devices where you think you are interfacing to a bunch of hardware for say your “Modem Chip”, “Audio chip”, “Comms chip” and even “Network chips”. In reality you are not talking to a bunch of hardware as you once would have done or now may do if you use FPGA’s, no you are waggling pins on an input status register or similar that is either polled, or causes an interupt on an old 8bit CPU that just runs crazy crazy fast…

At first you tend not to believe it, but when you do start to play with FPGA’s you start to realise. Compared to genuine logic macros on silicon FPGA macros are slow, as little as 1/10th the speed. Yet… even a “so so” FPGA design is five to ten times the speed of thes SoC chip interfaces.

We know how to make differential input XOR gates that run reliably at 50GHz clock speeds, you can if you have a large wadge of cash buy them from Digi-Key. The fact they are in a bonded out package tells you we can make them run faster on a chip therefore we could have say latches that run at oh about 25Ghz… Actually we do, you find them in test instruments and high frequency RF circuitry. You probably could not afford them because they are both “Mil Spec” and “Space Qialified” but for the right customers blowing tens of millions on a ride up, and ten times that again for a box that is going to float around in space for the next quater century, the technology is there.

One of the things I do is design prototype experimental payloads that use what is jokingly called “COTS” equipment. That is what is cheap throw away technology that is “Consumer Of The Shelf”, but is often cutting edge. Few realise that a modern Digital SLR and lenses you could buy in a shop or online for less than $1000 would be “against the rules” to send into space for a Low Earth Orbit”(LEO) because it’s resolution would be considerably better than one square Metre of earths surface per pixel. Oh and you could build it all into a Cube Sat 20cm by 10cm by 10cm… Most definately not what the USGov and similar want. However India has made it clear it’s not going to be bound by US rules, so have Japan and well both China and Russia, as for North Korea, they have yet to get into the commercial payload delivery game, but I suspect if we give them a little time…

As for “radio” links, a mobile phone with a few firmware tweaks could work quite comfortably in LEO. If you change the fractal antenna inside the consumer case to a neat little self deploying “origami folded antenna” three or more wavelengths in diameter…

You could build your own spy satellite for less than $3000 each. Oh and when you do the actual fairing payload calculations you could lift well upwards of 250 per launch[1] at quite a bit less than $20million per launch.

[1] If you were just talking lift mass then it’s well upwards of a thousand. The problem is getting them to spread out into longterm stable orbits where they won’t collide. You hear about SpaceX lifting 60 ~250Kg each desk sized Starlink satellites in a day. The limit of 60 is not constrained by lift or satellite technology issues, it’s just trying to get them into “StabO” that they have to sort out. At one point half of the Starlink Sats going up were deorbited and on their way down which is a shocking attrition rate. Only they are making six of those Starlink Sats a day and expect Starlink profits to be five times that of other Musk space projects by 2025. The satelites are launched into what they call “trains” which you used to be able to see them with the naked eye shortly after launch or a decent pair of binoculars as they got into orbit,

Because of the deploy limitations the Gen 2 Starlink Sats are bigger more robust and supposadly less visable,

Bob Paddock December 1, 2021 1:41 PM

@Peter A

“Cool technology museum… I would love to play around 🙂 Ok, ok, I know, it’s a closed research lab, sigh. … I personally saw all these VAXen etc. go to scrapyard… stupid bureaucracy…”

There is the
“Large Scale Systems Museum” in
New Kensington PA, just north of Pittsburgh PA.

It is a hands-on place where you are free to play.
Dave, the curator, loves DEC stuff which is what you find there mostly.
Still lots of Big Iron IBM too, and a few systems that even
Clive may never have heard of.

John December 1, 2021 2:53 PM


You make very good points!

I bypassed the whole blue pill thing and am shocked by how complex it is!! Even at such a low price.

I gotta believe that even given all the ‘protections’ blue pill resident code can be dumped by the chip manufacturer!

The whole Unix keep it simple and small message seems to be long gone which keeps small competition out!

This is the same old telephony game with a slightly different script.

A simple ham transceiver can ‘talk’ via a satellite. Why not texts? And as you point out why not direct to your cell ‘fone’? No big brother needed or wanted :).

All just SMOP [Small matter of programming] :).


vas pup December 1, 2021 5:30 PM

Scientists claim big advance in using DNA to store data

“DNA storage has a higher error rate than conventional hard drive storage. In collaboration with the University of Washington, GTRI researchers have come up with a way of identifying and correcting those errors.

!!!The work has been backed by the Intelligence Advanced Research Projects Activity (IARPA), which supports science that overcomes challenges relevant to the US intelligence community.”

Read the article for details!

vas pup December 1, 2021 5:46 PM

MI6 must adapt to new technology to survive, says spy chief

“Britain’s Secret Intelligence Service, better known as MI6, must become more open in order to do its secret work effectively. That’s according to its chief, Richard Moore, in his first major public speech since taking on the role in October 2020.

Artificial Intelligence (AI), quantum computing and digital technology have combined to completely transform the way human intelligence is gathered by spies, presenting MI6 with major challenges in the digital age.

Biometric data and facial recognition, for example, have made it far harder for intelligence officers to assume false identities in hostile countries without being recognized and exposed.

It was revealed in September that inside MI6 headquarters at Vauxhall Cross there is a real-life Q Section that works to provide Britain’s spies with the latest gadgets to keep them safe and secret.”

Peter A. December 2, 2021 3:29 AM

@Bob Paddock: thanks for the info. It is doubtful I’ll visit Pennsylvania in the foreseeable future, but I’m adding the museum to my bucket list.

Who? December 4, 2021 4:39 PM

Does Intel care about the security risks of that [not so] old computing technology? Then they should either support and patch all decade-old Intel ME releases or provide a ME firmware upgrade that effectively removes that misfeature from our computers.

Leave a comment


Allowed HTML <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre> Markdown Extra syntax via

Sidebar photo of Bruce Schneier by Joe MacInnis.