cross-posted from: https://lemmy.sdf.org/post/24645301

They emailed me a PDF. It opened fine with evince and looked like a simple doc at first. Then I clicked on a field in the form. Strangely, instead of simply populating the field with my text, a PDF note window popped up so my text entry went into a PDF note, which many viewers present as a sticky note icon.

If I were to fax this PDF, the PDF comments would just get lost. So to fill out the form I fed it to LaTeX and used the overpic pkg to write text wherever I choose. LaTeX rejected the file… could not handle this PDF. Then I used the file command to see what I am dealing with:

$ file signature_page.pdf
signature_page.pdf: Java serialization data, version 5

WTF is that? I know PDF supports JavaScript (shitty indeed). Is that what this is? “Java” is not JavaScript, so I’m baffled. Why is java in a PDF? (edit: explainer on java serialization, and some analysis)

My workaround was to use evince to print the PDF to PDF (using a PDF-building printer driver or whatever evince uses), then feed that into LaTeX. That worked.

My question is, how common is this? Is it going to become a mechanism to embed a tracking pixel like corporate assholes do with HTML email?

I probably need to change my habits. I know PDF docs can serve as carriers of copious malware anyway. Some people go to the extreme of creating a one-time use virtual machine with PDF viewer which then prints a PDF to a PDF before destroying the VM which is assumed to be compromised.

My temptation is to take a less tedious approach. E.g. something like:

$ firejail --net=none evince untrusted.pdf

I should be able to improve on that by doing something non-interactive. My first guess:

$ firejail --net=none gs -sDEVICE=pdfwrite -q -dFIXEDMEDIA -dSCALE=1 -o is_this_output_safe.pdf -- /usr/share/ghostscript/*/lib/viewpbm.ps untrusted_input.pdf

output:

Error: /invalidfileaccess in --file--
Operand stack:
   (untrusted_input.pdf)   (r)
Execution stack:
   %interp_exit   .runexec2   --nostringval--   --nostringval--   --nostringval--   2   %stopped_push   --nostringval--   --nostringval--   --nostringval--   false   1   %stopped_push   1990   1   3   %oparray_pop   1989   1   3   %oparray_pop   1977   1   3   %oparray_pop   1833   1   3   %oparray_pop   --nostringval--   %errorexec_pop   .runexec2   --nostringval--   --nostringval--   --nostringval--   2   %stopped_push   --nostringval--   --nostringval--   --nostringval--   %array_continue   --nostringval--
Dictionary stack:
   --dict:769/1123(ro)(G)--   --dict:0/20(G)--   --dict:87/200(L)--   --dict:0/20(L)--
Current allocation mode is local
Last OS error: Permission denied
Current file position is 10479
GPL Ghostscript 10.00.0: Unrecoverable error, exit code 1

What’s my problem? Better ideas? I would love it if attempts to reach the cloud could be trapped and recorded to a log file in the course of neutering the PDF.

(note: I also wonder what happens when Firefox opens this PDF considering Mozilla is happy to blindly execute whatever code it receives no matter the context.)

  • Troy
    38·
    20 days ago

    Your assertion that the document is malicious without any evidence is what I’m concerned about.

    At some point you have to decide to trust someone. The comment above gave you reason to trust that the document was in a standard, non-malicious format. But you outright rejected their advice in a hostile tone. You base your hostility on a youtube video.

    You should read the essay “on trusting trust” and then make a decision on whether you are going to participate in digital society or live under a bridge with a tinfoil hat.

    In Canada, and elsewhere, insurance companies know everything about you before you even apply, and it’s likely true elsewhere too. Even if they don’t have personally identifiable information, you’ll be in a data bucket with your neighbours, with risk profiles based on neighbourhood, items being insuring, claim rates for people with similar profiles, etc. Very likely every interaction you have with them has been going into a LLM even prior to the advent of ChatGPT, and they will have scored those interactions against a model.

    The personally identifiable information has largely been anonymized in these models. In Canada, for example, there are regulatory bodies like OSFI that they have to report to, and get audited by, to ensure the data is being used in compliance with regulations. Each company will have a compliance department tasked with making sure they’re adhering.

    But what you will end up doing instead is triggering fraudulent behaviour flags. There’s something called “address fraud”, where people go out of their way to disguise their location, because some lower risk address has better rates or whatever. When you do everything you can to scrub your location, this itself is a signal that you are operating as a highly paranoid individual and that might put you in a bucket. If you want to be the most invisible to them, you want to act like you’re in the median of all categories. Because any outlying behaviours further fingerprint you.

    Source: I have a direct connection to advanced analytics within insurance industry (one degree of separation).

    • BearOfaTime@lemm.ee
      52·
      20 days ago

      The personally identifiable information has largely been anonymized in these models

      This tells me all we need to know about you.

      You’re an apologist for these companies. It’s been repeatedly demonstrated such anonymization can be pretty easily reversed.

    • evenwicht@lemmy.sdf.orgOP
      3·
      20 days ago

      Your assertion that the document is malicious without any evidence is what I’m concerned about.

      I did not assert malice. I asked questions. I’m open to evidence proving or disproving malice.

      At some point you have to decide to trust someone. The comment above gave you reason to trust that the document was in a standard, non-malicious format. But you outright rejected their advice in a hostile tone. You base your hostility on a youtube video.

      There was too much uncertainty there to inspire trust. Getoffmylan had no idea why the data was organised as serialised java.

      You should read the essay “on trusting trust” and then make a decision on whether you are going to participate in digital society or live under a bridge with a tinfoil hat.

      I’ll need a more direct reference because that phrase gives copious references. Do you mean this study? Judging from the abstract:

      To what extent should one trust a statement that a program is free of Trojan horses? Perhaps it is more important to trust the people who wrote the software.

      I seem to have received software pretending to be a document. Trust would naturally not be a sensible reaction to that. In the infosec discipline we would be incompetent fools to loosely trust whatever comes at us. We make it a point to avoid trust and when trust cannot be avoided we seek justfiication for trust. We have a zero-trust principle. We also have the rule of leaste privilige which means not to extend trust/permissions where it’s not necessary for the mission. Why would I trust a PDF when I can take steps to access the PDF in a way that does not need excessive trust?

      The masses (security naive folks) operate in the reverse-- they trust by default and look for reasons to distrust. That’s not wise.

      In Canada, and elsewhere, insurance companies know everything about you before you even apply, and it’s likely true elsewhere too.

      When you move, how do they find out if you don’t tell them? Tracking would be one way.

      Privacy is about control. When you call it paranoia, the concept of agency has escaped you. If you have privacy, you can choose what you disclose. What would be good rationale for giving up control?

      Even if they don’t have personally identifiable information, you’ll be in a data bucket with your neighbours, with risk profiles based on neighbourhood, items being insuring, claim rates for people with similar profiles, etc. Very likely every interaction you have with them has been going into a LLM even prior to the advent of ChatGPT, and they will have scored those interactions against a model.

      If we assume that’s true, what do you gain by giving them more solid data to reinforce surreptitious snooping? You can’t control everything but It’s not in your interest to sacrifice control for nothing.

      But what you will end up doing instead is triggering fraudulent behaviour flags. There’s something called “address fraud”, where people go out of their way to disguise their location, because some lower risk address has better rates or whatever.

      Indeed for some types of insurance policies the insurer has a legitimate need to know where you reside. But that’s the insurer’s problem. This does not rationalize a consumer who recklessly feeds surreptitious surveillance. Street wise consumers protect themselves of surveillance. Of course they can (and should) disclose their new address if they move via proper channels.

      Why? Because someone might take a vacation somewhere and interact from another state. How long is a vacation? It’s for the consumer to declare where they intend to live, e.g. via “declaration of domicile”. Insurance companies will harrass people if their intel has an inconsistency. Where is that trust you were talking about? There is no reciprocity here.

      When you do everything you can to scrub your location, this itself is a signal that you are operating as a highly paranoid individual and that might put you in a bucket.

      Sure, you could end up in that bucket if you are in a strong minority of street wise consumers. If the insurer wants to waste their time chasing false positives, the time waste is on them. I would rather laugh at that than join the street unwise club that makes the street wise consumers stand out more.